HIL-SERL tutorial for simulation fixed and expanded

[spirosperos]

What this does

This PR fixes two critical bugs in the HIL-SERL simulation training framework and adds comprehensive documentation:

Bug Fixes:

• 🐛 Bug Fix: Fixed control_time_s parameter not being respected during training - episodes were always 10 seconds regardless of configuration
• 🐛 Bug Fix: Fixed random_block_position flag not being properly passed to the gym environment, preventing cube randomization

Improvements:

• 📚 Documentation: Added comprehensive training guide (hil_serl_simulation_training_guide_README.md) for HIL-SERL simulation training
• 🔧 Enhancement: Added extensive debug logging for easier training monitoring and troubleshooting

Key Changes:

• Implemented TimeLimitWrapper class to properly enforce episode time limits based on control_time_s configuration
• Added random_block_position parameter to environment configuration and properly passed it to gym environment
• Enhanced logging throughout training process with detailed episode progress, time tracking, and environment state information

How it was tested

• Time Limit Fix: Verified that setting "control_time_s": 40.0 in configuration now properly limits episodes to 40 seconds instead of default 10 seconds
• Cube Randomization Fix: Confirmed that "random_block_position": true now properly randomizes cube positions between episodes
• Debug Logging: Tested debug outputs during training to ensure proper monitoring of episode progress, time remaining, and environment state
• Documentation: Verified all commands and configurations in the new README work correctly with the fixed implementation

Test Commands:

# Test recording with new time limit
python -m lerobot.scripts.rl.gym_manipulator --config_path examples/hil_serl_simulation_training/hi_rl_test_gamepad.json

# Test training with both fixes
python -m lerobot.scripts.rl.learner --config_path examples/hil_serl_simulation_training/train_gym_hil_env_gamepad.json
python -m lerobot.scripts.rl.actor --config_path examples/hil_serl_simulation_training/train_gym_hil_env_gamepad.json

How to checkout & try? (for the reviewer)

1. Test Time Limit Fix:

# Modify control_time_s in hi_rl_test_gamepad.json to 20.0 and verify episodes last 20 seconds
python -m lerobot.scripts.rl.gym_manipulator --config_path examples/hil_serl_simulation_training/hi_rl_test_gamepad.json

2. Test Cube Randomization:

# Set random_block_position to false in config and verify cube stays in same position
# Set to true and verify cube randomizes between episodes
python -m lerobot.scripts.rl.gym_manipulator --config_path examples/hil_serl_simulation_training/hi_rl_test_gamepad.json

3. Test Training with Both Fixes:

# Terminal 1: Start learner
python -m lerobot.scripts.rl.learner --config_path examples/hil_serl_simulation_training/train_gym_hil_env_gamepad.json

# Terminal 2: Start actor  
python -m lerobot.scripts.rl.actor --config_path examples/hil_serl_simulation_training/train_gym_hil_env_gamepad.json

4. Review Documentation:

# Check the new training guide
cat examples/hil_serl_simulation_training/hil_serl_simulation_training_guide_README.md

Expected Behavior:

• Episodes should respect the control_time_s setting (30s in training config, 40s in recording config)
• Cube should randomize position when random_block_position: true
• Debug logs should show detailed episode progress and time tracking

[lukicdarkoo]

There is a mess up with the branches, fixed in #2