Merge remote-tracking branch 'origin/main'

# Conflicts:
#	rlgym_tools/version.py

Update version.py

Update version.py

Bump version to 2.6.2 and update release notes with new features

v2.6.1

- Update carball to support latest replay format changes
- Update convert.py to use new flip reset packet data
- Change ReplayMutator to use .npy files instead of .dat, and produce new replay frames.
- Add scripts to make, visualize, and validate replay frame files.

v2.6.0

- Replace TeamSpiritRewardWrapper with more configurable DistributeRewardsWrapper
- Fix RewardReport not tracking episode ends automatically

v2.5.0

- VelocityPlayerToBallReward is now normalized to give 1 reward per second at max speed instead of 1 per step.
- Refactor BallTravelReward:
  - All weights now default to 0.
  - Last_touch_agent is now a list called last_touch_agents.
  - Rewards are averaged between agents touching last step and agents touching this step.

- Split VelocityPlayerToBallReward into different reward functions in…

…stead of using parameters:

  VelocityPlayerToBallReward, DotQuotientVPBReward, TrajectoryComparisonVPBReward
- Redo parameters in BallTravelReward to make them clearer

Fix reward calculation for GoalProbReward when gamma<1

Update carball