Add selective component analysis to NLP processor

[jwalsh]

Staff Engineer Review (Alex Rivera)

Currently, the NLP processor forces computation of all analysis components (entities, tokens, sentences, POS tags, dependencies) even when a user might only need one specific type of analysis. This is inefficient and leads to unnecessary processing overhead.

Requirements

• Add ability to specify which analysis components to run
• Allow users to disable unneeded pipeline components for better performance
• Update API to support this selective analysis

Implementation Notes

• Modify analyze_text() method to accept component selection parameters
• Update server API to pass these parameters to the processor
• Document new parameters in API docs