[ML] Enhancements to model memory limit calculator

[peteharverson]

Meta issue outlining the enhancements required for the anomaly detection job model memory limit calculator for 7.6:

• Add in functional test coverage for the existing /api/ml/validate/calculate_model_memory_limit endpoint, using the ecommerce and farequote archives currently available for the ML functional tests.

  • Range of metric field names (0 - 5 ?)
  • Range of influencers (0 - 5 ?)
  • Use of a query which limits the number of partition / influencer field values used in the calculation.

• Edit the multi metric job creator to ensure count detectors with a partition field get included in the calculation. These currently do not affect the calculations as only detectors which use a metric field name are included.

• The MML estimate isn't re-calculated in the multi-metric wizard when influencers are added or removed. The endpoint should be called when the influencers are changed.

[elasticmachine]

Pinging @elastic/ml-ui (:ml)

[darnautov]

Closed by #54557 and #54573