Verify the number of tokens for the unified gap analysis prompt to ensure context is not overloaded #5949
Description
This was tangentially brought up by @jcscottiii for another project.
It would be wise to ensure we are not overloading the 1M token count context for Gemini before we try and submit the prompt for response. We can do this with the countTokens method. If the prompt is too large, just revert to using the typical 3-prompt flow.