Enable model switching

[remlaps]

Update Thoth to be able to switch between LLM models when network disruptions or other challenges are encountered. Example cases:

1. Gemini Pro is returning 503 errors, so switch automatically to using Gemini Flash.
2. A free API key has reached its quota, so auto-switch to use a paid API key.
3. Gemini is totally unreachable, so switch over to using ArliAI models (or vice versa).

[remlaps]

Implemented option 1 in this branch - https://github.com/remlaps/Thoth/tree/MultipleModels.

Will test until we see it succeed a few times before merging. It should now try a series of models if it gets 429 or 503 errors. Options 2 and 3 have not been implemented yet.

[remlaps]

ChatGPT set-up some integration testing for me, all tests executed, and merged into master. Will observe for days/weeks to see what happens when 429/503 errors are encountered.

Here is the PR: #53

And here are the test results:

All core functionality validated:

Model switching through fallback list
Dry-run mode (marks but doesn't switch)
No duplicate tracking
Reset functionality
Single model handling
Correct ordering of models used
Whitespace trimming in config parsing```

===

`
=== Test: Basic model switching ===
  + Switched through all models
  + Final rate_limited: ['model-1', 'model-2', 'model-3']
  + get_models_used(): ['model-1', 'model-2', 'model-3']
  PASSED

=== Test: Dry-run mode ===
  + Dry-run marked model: ['model-a']
  + Still on original model: model-b
  + Real switch worked afterward
  PASSED

=== Test: No duplicates in rate_limited ===
  + No duplicates: ['model-x', 'model-y']
  PASSED

=== Test: Reset ===
  + Reset successful
  + Current: a, Rate-limited: []
  PASSED

=== Test: Single model (no fallback) ===
  + Single model handled correctly
  PASSED

=== Test: get_models_used() ordering ===
  + Models in correct order: ['first', 'second', 'third']
  PASSED

=== Test: Whitespace handling ===
  + Whitespace stripped correctly
  + Models: ['model-1', 'model-2', 'model-3']
  PASSED

============================================================
+ ALL INTEGRATION TESTS PASSED
============================================================`

[remlaps]

Related: Enable split LLM usage between Intro post and summary replies