Ragatouille

(notes encompassing changes in the last few PyPi releases that were undocumented until now)

**Changes**:
- Query only a subset documents based on doc ids by PrimoUomo89 https://github.com/bclavie/RAGatouille/pull/94
- Return chunk ids in results thanks to PrimoUomo89 https://github.com/bclavie/RAGatouille/pull/125
- Lower kmeans iterations when not necessary to run more https://github.com/bclavie/RAGatouille/pull/129
- Properly license the library as Apache-2 on PyPi

**Fixes**:
- Dynamically increase search hyper parameters for large k values and lower doc counts, reducing the number of situations where the total number of documents return is substantially below `k` https://github.com/bclavie/RAGatouille/pull/131
- Fix enabling the use of Training data processing with hard negatives turned off by corrius https://github.com/bclavie/RAGatouille/pull/117
- Proper handling of different input types when pre-processing training triplets by GautamR-Samagra https://github.com/bclavie/RAGatouille/pull/115

Minor fixes&improvements release.

Community contribs:
- https://github.com/bclavie/RAGatouille/issues/103 fixes **properly** pushed to main thanks to bjsi
- Better verbosity handling https://github.com/bclavie/RAGatouille/pull/112 by Potrock
- Metadata support in Langchain integration https://github.com/bclavie/RAGatouille/pull/104 by GMartin-dev

- Fix newly introduced dependency issue

- Fixes sometimes skipped shuffling of training triplets
- Fixes accidental duplicates when input training data has many more positives than negatives.
- Bump to `colbert-ai` 0.2.18, fully removing multiprocessing calls when indexing

Fixes & minor improvements:
- Better verbosity control, especially for high-CRUD encode() scenarios
- Fixed max document length being often set too low when using in-memory encode() or rerank()
- Allow forced overwrite of indexes (https://github.com/bclavie/RAGatouille/issues/63)
- Fix wrong argument being passed to negative miner (should not have had any impact in practice)

Initial "proper" release. Changelogs begin now.