Latium starting resources fail

[elexis]

On most generations of Latium, we get a warning about starting resources having failed to place.

Beside the warning being annoying, the missing mines are quite bad.

[elexis]

It's because cliffs are painted, then players placed inside it and the area flattened, but the original area (clCliff) is still attempted to be ignored.

[mimo]

In addition, with the commandx.txt attached and r21034, there are fishes on land.

As both cases i've noticed on this map are on the lateral borders, maybe the map rotation was not properly applied?

[elexis]

In 21069:

rmgen random placement and Entity instantiation refactoring, fixes #4992.

ChainPlacer, ClumpPlacer, SimpleObject receive the vectors that are in place everywhere already, refs #4845.
Add public setCenterPosition to CenteredPlacer and Group rather than writing private properties of the prototypes.
ChainPlacer and ClumpPlacer simplifications, deduplication and renames, refs #4805.

Replace placeObject global with RandomMap placeEntity members, refs #4804.
Split to placeEntityPassable / placeEntityAnywhere, as well as validTilePassable / validTileAnywhere to more cleanly distinguish actor and casual entity placement.
No more does SimpleObject create Entity instances and register entityIDs if they are never placed.
Removes the map global reference from the Entity constructor, refs #4964.
By definition of what is passed to the engine, an Entity has an ID and position, so keep it impossible to create Entities without IDs.

Implement randomPositionOnTile so that there aren't different implementations thereof, including unintented ones as in ardennes_forest.js in rP21021.
On Caledonian Meadows, remove unused pathplacing code, to be superseded by #4368.
On Schwarzwald, delete unused startLocations following rP20864.
On Latium, replace complicated duplicated hardcoded fish location computation with a simple HeightConstraint, refs #4960.

[OptimusShepard]

If the player starts on the map pompeii on the half island, this error occurs too. The mapsize was choosen tiny.

[elexis]

The abandoned WIP branch I had been working on in november. Don't know what it contains.

[elexis]

(Oh, this branch is even from 2016, so it's not the one latium rewrite I had been looking for. I went through the list of diffs in there and saw most of it being committed and few cleanup I'm about to commit. Searched all files for "latium" on my disk but can't find the rewrite. Must have been abandoned and deleted to be never seen again.)

[temple]

In 21552:

Fix a bug in Latium and Ardennes Forest

Differential Revision: ​https://code.wildfiregames.com/D1387
Reviewed by: elexis
Refs: #4960

[elexis]

Missing resources on tiny maps can occur on many maps for different reasons.

Primarily placePlayerBaseMines and placePlayerBaseChicken should not receive distance but min and max distance, similar to placePlayerBaseTrees. Perhaps that can still be realized for this release.

Perhaps we can replace the placePlayerBase functions entirely with fewer new library functions that share some of the properties of the placePlayerBaseFunctions, such as retry loops, radial distribution similar to createObjectGroup (Inari also proposed a CircleConstraint with a min and max radius recently, that might be utilized). (That certainly won't be done this release)

[elexis]

not so fast

[elexis]

meh