Refactored AnthropicAugmentedLLM and GoogleNativeAugmentedLLM

[janspoerer]

No description provided.

[evalstate]

@janspoerer -- is this ready (i assume the other PR relates to this). if so, i'd quite like to run it through the e2e suite. how would you describe the main goal of this refactoring btw?

[janspoerer]

This is ready for testing. But one could also test #311 as a whole when it is ready, as it builds on this PR.

How much does one full e2e run cost in credits?

The main goal is to split the code into smaller chunks, especially the completion methods (_anthropic_completion()), which are now a bit more legible/concise and maintainable (or so I hope).

This was with the truncation feature (#311) in mind, which builds on the commits from this branch. I wanted to insert the truncation feature and noticed that it would be easier if the provider methods had methods that were a bit shorter.