fix: Use entire file as context in "Optimize Code" prompt

[bryanhuhta]

✨ Description

Currently, we only use the snippet of code displayed to the user as context for the LLM prompt. This has the downside of making the LLM reason with less information. If we pass the entire file (if we have it) to the LLM prompt, this should improve its reasoning ability and give users the chance to ask follow up questions about portions of the code which might be in the same file, but not in the profiling snippet.

Here's an example of a short code snippet in a large file:

When asked about code in a different part of the same file, the LLM reports it can only reason about the current snippet:

With the changes in this PR, the LLM can now reason about the the entire file. Here's a snippet that's part of a larger file:

And here is the LLM's response to asking about a specific part of the file not found in the snippet:

🧪 How to test?

The unit tests should pass

[github-actions]

Unit test coverage

Lines	Statements	Branches	Functions
	12.52% (554/4423)	8.96% (137/1529)	9.23% (122/1321)