Create an action delay range

[jgillick]

Note

Adds configurable action delay (fixed int or per-env random range) via a ring buffer in BaseActionManager, integrated into position action managers.

• Action Management
  • Support delay_step as int or (min, max); validate inputs and treat 0 as no delay.
  • Implement ring buffer–based delay with per-environment indices via _apply_action_delay; step() now applies delayed actions.
  • Initialize delay buffers and env indices in build(); assign per-env random delays in reset().
• Position Managers
  • Update delay_step type to accept range and call super().build() to initialize delay buffers.

^{Written by Cursor Bugbot for commit 7ec2608. This will update automatically on new commits. Configure here.}