FlxAnimationController: fix off-by-one and negative modulo bugs

[dtwotwo]

fixes bugs in FixAnimationController

• corrected off-by-one error in add() frame validation
  (while (i-- >= 0) -> while (i-- > 0))
• fixed negative frame index handling in set_frameIndex:
  negative values are now clamped to 0 with a log warning,
  values >= numFrames are wrapped with a log warning

[Geokureli]

For posterity, the off-by-one bug was introduced here: #2762 (D'oh!)

Can you provide examples that would have failed in prior versions? We should add unit tests for them. I'm also not sure I understand how negative frame indices are used in set_frameIndex

[dtwotwo]

For posterity, the off-by-one bug was introduced here: #2762 (D'oh!)

Can you provide examples that would have failed in prior versions? We should add unit tests for them. I'm also not sure I understand how negative frame indices are used in set_frameIndex

off-by-one in add():

the loop while (i-- >= 0) fires an extra iteration on every add() call. when i is 0, the condition 0 >= 0 is true, i becomes -1, and the body accesses framesToAdd[-1]. on cpp this is an out-of-bounds read that corrupts the array. The same file already uses the correct pattern in removeInvalidFrames - while (i-- > 0).

negative modulo in set_frameIndex:

frameIndex is a public property, and internally copyFrom can pass the default value of -1 into it. in Haxe, -1 % 4 == -1, not 3, so _sprite.frames.frames[-1] is an out-of-bounds access. FlxMath.mod handles wrapping correctly.

unit tests:

added testAddAllValidFrames, testAddSomeInvalidFrames, testAddSingleValidFrame, testAddAllInvalidFrames, testFrameIndexNegativeWraps for test possible fallbacks

• CI on old code without fix (cpp fails): https://github.com/noegarbage/flixel-prs/actions/runs/21875412703 (tests used wrong sprite sheet size, but failures confirm the same bugs)
• CI on new code with fix: https://github.com/noegarbage/flixel-prs/actions/runs/21878869711

[Geokureli]

Thanks!