Fix calibration adjustments to move immediately

[InertiaImpact]

This pull request refactors how fine-tuning adjustments are sent to hardware modules, ensuring that after a correction is applied, the module immediately moves to the new position for user feedback. It introduces a helper function for building adjustment commands, updates the backend to use this helper, removes redundant client-side requests, and adds tests for the new logic. User-facing descriptions have also been updated for clarity.

Backend improvements:

• Added a new helper function build_tuning_adjust_commands in tuning.py to generate commands that both save a tuning adjustment and immediately preview it by moving the module.
• Updated auto_tune_route in server/app.py to use the new helper, ensuring that after an adjustment, the module is moved immediately for feedback. The response now also indicates that the adjustment was previewed.
• Added unit tests in tests/test_tuning.py to verify that build_tuning_adjust_commands generates the correct commands and properly validates input.

Frontend changes:

• Removed unnecessary client-side requests in app.js that previously re-sent the same character index after an adjustment, since the backend now handles immediate previewing. [1] [2] [3]

User interface updates:

• Updated instructional text in index.html to clarify that corrections now move modules immediately, and that negative corrections may require a full forward revolution. [1] [2]

[csader]

Clean, well-scoped fix. The extraction into tuning.py with validation and tests is solid, and removing the redundant client-side goto_char requests makes sense now that the backend previews immediately.

One question before I merge: can you confirm the g command (m{id}g{position}) is "goto absolute step position"? The old code was sending goto_char (character index) client-side — the new approach sends a step position directly, which is semantically different. Just want to make sure the firmware handles g as an absolute step goto rather than a character index.

Also noting there's no upper-bound validation on step_position (module_id caps at 254, char_index at 63, but step_position only rejects negatives). Not blocking — firmware should reject out-of-range values — but worth considering for consistency.

Looks good to merge once confirmed.

[InertiaImpact]

As for the g command - it's referenced here as

Moves the reel to an absolute step position (0 to totalStepsPerRev − 1). Sets the current flap index to unknown.

• after saving the tuned position, we move directly to the adjusted raw step so the compensation is visibly previewed, even if the logical flap index hasn’t changed.

I also added upper-bound validation so the step must be within the module’s calibrated revolution range: 0 <= step_position < calibration_steps.

[csader]

Clean, focused change. The g command approach is the right call — re-sending the character index could be ignored since firmware still considers that flap active.

Extraction into tuning.py with validation and tests is solid. Ready to merge.