Bug fix: Keep generated resource tile state valid across recompute redraws

[instafluff0]

This is a fix for som_rybov's bug report, here, save attached.

In a nutshell, I traced the crash back to Tile_has_colony getting a NULL tile. However patching that (bailing if NULL) fixed the symptom but not the underlying problem.

As best I can tell, the crash is from Trade_Net::recompute_resources receiving a bad tile pointer while we have the temporarily expanded Map.TileCount to include generated resource tiles.

1. patch_Trade_Net_recompute_resources builds is->resource_tiles.
2. It saves the real tile count:

is->saved_tile_count = p_bic_data->Map.TileCount;
p_bic_data->Map.TileCount += is->count_resource_tiles;

3. Trade_Net::recompute_resources loops over 0 .. Map.TileCount - 1.
4. The patched Map_get_tile_when_recomputing_resources_* functions are supposed to return:

• real map tile if index < saved_tile_count
• generated resource tile otherwise

5. But Tile::has_colony(NULL) means (I think) the first tile lookup returned NULL.

So if a resource recompute happens while saved_tile_count is already active, then saved_tile_count can be overwritten with an already-inflated tile count. Then an artificial tile index is misclassified as a real map index, so the wrapper calls real Map_get_tile with an out-of-range index. That returns NULL.

The saved game moves to the next turn after these changes, and calculated resources appear correct afterward.

Also: I noticed in the course of debugging this we had duplicate define Tile_has_colony entries in civ_prog_objects.csv and removed the 2nd one.

PS - holy moly, the repo is really heating up with PRs. I hope all is well and this isn't all too much too quickly.

bug.zip

[maxpetul]

Thanks for the fix. This all looks fine except for the deletion of line 24930, why'd you do that? Also seeing this reminds me I wanted to combine all those patch_Map_get_tile... functions. There's no reason to have separate patch functions if they're all the same.