Codechange: Use std::find_if to find or assign a text effect slot.

This replaces an index-based loop.

Codechange: Use std::array for TemporaryStorageArray.

This removes the need for initialisation with memset().

Codechange: Use std::array and std::unique_ptr for PersistentStorageArrays.

This (mostly) avoids the need for manual memory management and copying.

Codechange: Using alias and std::array for company expense storage. (#11273)

This simplifies passing yearly expenses to functions and use of std algorithms.

Fix: asserts unintentially being partially disabled in release-builds (#11268)

The nlohmann-json header file includes assert.h, which rudely resets
the assert macro to what that header thinks is right. As we set the
assert macro to be active with release builds when WITH_ASSERT is
active, this means that every file including nlohmann-json has their
asserts disabled (for release-builds) but files that don't do no.

Let's avoid this issue, by telling nlohmann to not include assert.h.

Fix: Inaccurate waiting cargo total in station window when using cargodist (#11213)

For stations with many flows and/or small cargo packets,
due to accumulated inaccuracies in DivideApprox.

The displayed total should match GoodsEntry::TotalCount().

Update: Translations from eints
english (us): 4 changes by 2TallTyler

Update: Translations from eints
french: 2 changes by ottdfevr

Codechange: Return update nwid_begin instead of count.

This avoids needing to keep track of how many widget parts have been consumed, instead we only to ensure nwid_begin < nwid_end.

Codechange: Use begin/end of nwidget parts of begin/length.

This simplifies processing nwidget parts as, unlike the remaining length, the pointer to the end of the list never changes. This is the same principle as we use(d) for tracking end instead of length for C-style strings.

And this removes 160~ instances of the lengthof() macro.

Codechange: Allow using more than 65536 NewGRF string IDs.

NewGRF string allocation allowed up to 524288 strings, however stringid was passed as uint16_t which limits to 2^16.

Change: Replace fixed length _grf_text array with vector.

Additionally reshuffle GRFTextEntry for better alignment.

This removes a mostly-unused static 20MB allocation.

Update: Translations from eints
english (au): 2 changes by krysclarke
russian: 3 changes by Ln-Wolf
finnish: 2 changes by hpiirai
portuguese (brazilian): 2 changes by pasantoro

Codechange: Reorder some high-use structs to reduce their size. (#11201)

This reduces GoodsEntry from 144 to 136 bytes (thereby reducing Station from 9704 bytes to 9192 bytes), and CargoPacket from 40 bytes to 32 bytes.

Codechange: workaround CMake/Xcode duplicate file name issue (#11186)

Having a library with files with the same name isn't supported in CMake's Xcode project file generation: https://gitlab.kitware.com/cmake/cmake/-/issues/20501. One of the files is renamed to work around this bug.

Update: Translations from eints
dutch: 1 change by Afoklala

Codechange: add tests for GS <-> AdminPort JSON conversion (#11252)

While at it, fix a bug where booleans were made integers.

Fix: [Emscripten] config not saved on exit game (#11248)

When changing a Game Option and pressing Exit Game, the changes
were not actually stored. This because the post-mainloop code
was never executed for Emscripten.

Add: [CI] Create and store breakpad symbols for releases (#11247)

Additionally, also store PDB and exe files on the Symbol Server
for easier debugging with MSVC.

Update: Translations from eints
catalan: 2 changes by J0anJosep

Add: calendar date for Survey results

This means no heuristics is possible on around which date people
play the game.

Change: [Win32] don't allocate 192KiB of memory on the stack on crash (#11240)

Heap is out of the question, as it might be corrupted.
Allocating this much on stack is silly.

So instead, allocate virtual pages to write the information in.

Change: mention in MacOS / Windows crashlog popup when files couldn't be generated (#11239)

For example, if the crash.sav couldn't be created, and the TryExcept
detected an issue, write out that the file is not available in
the popup.

Codechange: [MacOS] use backtrace() instead of our custom variant (#11233)

As mentioned in the comment, we only did it ourselves as we once
were compatible with versions before 10.5. But that time has long
gone. So let's update the code to a bit more modern approach.

Update: Translations from eints
english (au): 1 change by krysclarke
finnish: 3 changes by hpiirai
danish: 1 change by bscargo
french: 1 change by glx22
portuguese (brazilian): 1 change by pasantoro

Change: [Win32] unify the crashlog handler with the other OSes (#11236)

While at it, make the crash text a bit more readable, and sync
this with MacOS.

Update: Translations from eints
russian: 1 change by Ln-Wolf
turkish: 1 change by densxd
portuguese: 1 change by azulcosta
polish: 3 changes by pAter-exe

Update: Translations from eints
catalan: 30 changes by J0anJosep

Update: Translations from eints
italian: 43 changes by Rivarossi
french: 1 change by ottdfevr

Update: Translations from eints
polish: 2 changes by pAter-exe

Codechange: [Win32] unify AppendDecodedStacktrace and LogStacktrace (#11220)

AppendDecodedStacktrace was already doing the same as LogStacktrace,
just with a different name.

Update: Translations from eints
turkish: 2 changes by BeratSJ

Remove: [Win32] module-list from crash.log (#11219)

Only Windows implemented this, and it opens the files to read them
to get a CRC. Doing this in a crash-handler is strange at best.

Lastly, nobody has actually ever used this information to come to
some sort of conclusion. The module-list is used in combination
with the crash.dmp, but this information is already embedded in
there.

Add: use breakpad to create crash.dmp on MacOS / Linux too (#11202)

Normally only the Windows platform could create a crash.dmp, making
analysing crash-reports from MacOS / Linux rather tricky.

Remove: [Win32] (pointer-only) stack trace in crash.log (#11211)

It only contains pointers, which nobody can decipher anyway.
So instead, just report "Not supported", like other targets do
when they can't print a sane stack trace.

Change: Remove autosave from settings window (#11218)

It is broken since 4f4810d and it's also in game options.

Update: Translations from eints
german: 40 changes by Wuzzy2

Remove: last remnants of SunOS (#11210)

We have no evidence SunOS has been tried in the last few years,
and the code is awkward enough to just remove.

Update: Translations from eints
dutch: 2 changes by Afoklala

Fix: [MacOS] screen looks blue-ish when using newer SDKs (#11207)

The define kCGBitmapByteOrder32Host changed (around SDK 12?)
into an enum, which means an old #ifndef was triggering,
overwriting the value to 0. Sadly, 0 means Order16Big, causing
RGBA to become GRAB, which results in strange colours.

As we no longer support PPC, drop that piece of code completely.

Update: Translations from eints
swedish: 1 change by joeax910

Update: Translations from eints
dutch: 1 change by Bouke

Codechange: make explicit when a TileIndex is cast to its basetype (#11190)

This prevents people accidentially assigning a TileIndex to a Date
or any other type they shouldn't.

Update: Translations from eints
swedish: 1 change by Brickblock1
english (us): 2 changes by 2TallTyler

Update: Translations from eints
english (au): 2 changes by krysclarke
french: 1 change by Lishouuu
portuguese (brazilian): 2 changes by pasantoro

Update: Translations from eints
korean: 3 changes by telk5093
russian: 2 changes by Ln-Wolf
portuguese: 3 changes by azulcosta

Codechange: introduce week/quarter triggers for TimerGameCalendar (#11189)

This simplifies code that triggers on these periods.

Codechange: simplify the type of loaded_at_xy / next_station (#11182)

It was TileOrStationID, most likely to make sure both types in
the union are identical. But as TileIndex is a StrongTypeDef
that becomes a bit weird. So instead, still make sure they are
of equal size, but define their individual types better.

Codechange: make no assumptions on how the internals of TileIndex work (#11183)

Basically, avoid ".value", and just cast it to its original type
if you want to retrieve this.