MPC differential tuning bug fixes

[tombrazier]

Description

This PR fixes several problems with the MPC differential tuning method.

• Moved option to enable MPC debugging to Configuration.h.
• Fixed derivative calculation (was delta temp * delta time, should have been delta temp / delta time, worked anyway because delta time happened to be 1).
• Fixed time calculation for when maximum rate is observed. The return value of get_elapsed_heating_time() was not yet initialised at the point where it was being called.
• Fixed call to tuner.get_time_fastest()) which should have been tuner.get_temp_fastest()). (See equation 17 at https://marlinfw.org/docs/features/model_predictive_control.html.)
• Added a second check for nonsensical MPC_BLOCK_HEAT_CAPACITY and MPC_SENSOR_RESPONSIVENESS values after the "Refining estimates..." step. This could happen when using the automatic tuning method which attempts an asymptotic tune and then falls back to differential if the results don't make sense. See [BUG] M306 T produces negative responsiveness value #27181.

Requirements

Enable MPCTEMP.

Benefits

Differential tuning now works and it is more likely that the automatic tuning method will fall back to differential tuning when needed. In particular this should make M306 T work seamlessly for Revo hotends.

Configurations

Enable MPCTEMP.

Related Issues

• [BUG] M306 T produces negative responsiveness value #27181
• [BUG] Wrong MPC settings can cause thermal runaway #27253

[tombrazier]

Build error related to FT_MOTION, not this PR.

[thisiskeithb]

Build error related to FT_MOTION, not this PR.

Breaking changes were reverted in ade05c0, so merge in the latest bugfix-2.1.x or rebase your PR & force-push changes.

[thinkyhead]

Breaking changes were reverted…

NOTE: A PR doesn't need to be updated if its changes don't overlap with changes to bugfix-2.1.x. The CI tests are done by merging the changes from the PR onto the current bugfix-2.1.x at the time the tests are initiated. If the bugfix-2.1.x branch has a compile error and then it is fixed, you don't have to take extra steps to update your branch if there are no conflicting changes; just re-run the tests.

[thisiskeithb]

just re-run the tests.

That only works if you have proper repo permissions, so merging in the latest changes works universally for everyone.

[thinkyhead]

Hmm, where did we put that "re-run yer own bloody PR's tests" permission? That seems like something we should have.