Generation 1 Disable handling requires tracking the move in addition to the slot

[scheibo]

Describe the bug:

In Generation I, the move which is disabled (wPlayerDisabledMoveNumber) needs to be tracked in addition to its index/duration (wPlayerDisabledMove).

Example:

This is specifically relevant wrt Transform. Currently the engine checks whether a move is disabled by comparing the move at the disabled index to the move trying to be used. This will actually handle the Mimic case, because if eg Tackle in slot 2 gets disabled but Mimic in slot 1 is replaced by Tackle, the cart and the engine will allow the slot 1 Tackle to be selected by will not let it be executed. This is not the case with Transform. If a Pokémon's slot 2 tackle gets Disabled and then Transforms, currently the engine will not only correctly ban the Transformed Pokemon from selecting what is at slot 2 (say eg. Water Gun), it will erroneously not prevent the mon from now using the Transformed pokemons slot 1 Tackle, it will also ban the pokemon from using Mimic and then a Mimicked Water Gun

Expected behavior:

Pokemon is not able to use the slot that was disabled or the move corresponding with at the slot that was disabled for the duration of disable

Additional context:

https://discord.com/channels/689316505560809576/689316506265321535/1481050218223894529

Thanks to @andrebastosdias for discovering this

[scheibo]

Solution sketch:

- state: u16 = 0,
+ state: u12 = 0,
+ /// The identity of whom the active Pokémon is transformed into.
+ transform: u4 = 0,
  /// The remaining HP of the Substitute.
  substitute: u8 = 0,
- /// The identity of whom the active Pokémon is transformed into.
- transform: u4 = 0,
+ /// The identity of whom the active Pokémon's disabled move originated from.
+ disable_source: u4 = 0,
  /// The remaining turns the move is disabled.
  disable_duration: u4 = 0,
  /// The move slot (1-4) that is disabled.
  disable_move: u3 = 0,
  /// The number of turns toxic damage has been accumulating.
  toxic: u5 = 0,

Its obviously safe to remove the 4 bits from accuracy as it only uses a u8. Removing 4 bits from Bide seems like it would result in correctness issues since on the cart it can go up to 65535 and will potentially roll over... except no Pokémon can sustain more than 702 damage (Chansey has 703HP) to be able to return the hit, so we technically only need 10 bits for Bide.

Tasks:

• add failing test case for this specific edge condition
• update data and code
• change offsets and JS package to handle this logic. debug logic and documentation needs to change as well
• update min bits doc for bide/disabed & percentages
• migrate data in old test cases to use new data protocol so that they still test the same thing post change (use serde.zig and both the old and new definitions of gen1/data.zig). need to make sure existing .bin files dont have disable volatiles set or else need to figure out disable_source argument