[FEATURE] Better handle the behavior diff for semantic highlight sing vs batch

Is your feature request related to a problem?

We have behavior diff between the semantic highlight using standard Highlight framework vs System generated search processor approach:

Standard Highlighter Framework:

• Cannot support non-string fields
• Fails early when encountering lists or objects - returns null and silently skips highlighting (Need to double confirm)

Batch Inference Processor Approach:

• Can support non-string fields
• Can handle lists and objects successfully (Need to double confirm)

The User Experience Gap:

Currently, when a user tries to highlight a field that's a list:

The standard approach silently returns null and skips highlighting (no feedback to user) The processor approach can actually highlight the list successfully

What solution would you like?

Rather than silently skipping, we should inform users when they attempt to highlight a list or object field using the standard approach. Should we try to throw an error(may not be easy to do this since we may fail in core before we get into the neural plugin) to let user know if they want to highlight non string field they can try the processor approach? Or we simply update the documentation to make this clear and let user figure it out by themselves?

What alternatives have you considered?

A clear and concise description of any alternative solutions or features you've considered.

Do you have any additional context?

Add any other context or screenshots about the feature request here.