fix(ColbertRerank): calculate ColBERT similarity per token rather than vs pooled query embeds (#11335)

fix: calculate ColBERT similarity per token rather than vs pooled query embedding

* Generate skeleton for Groq LLM integration.

* Use OpenAI-like to implement Groq LLM integraton.

Co-authored-by: gavinsherry <gavin.sherry@gmail.com>

---------

Co-authored-by: gavinsherry <gavin.sherry@gmail.com>

* add system prompt workaround

* remove workaround flag

* fix lint

* chore: fix doc comments

* Update base.py

* Update base.py

Update customization.rst

* fixed link to LlamaHub
* updated package name to reflect what's in PyPi

cr

Bump aiohttp from 3.9.1 to 3.9.2 in /llama-index-integrations/embeddings/llama-index-embeddings-fireworks (#11297)

Bump cryptography from 42.0.2 to 42.0.4 in /llama-index-integrations/embeddings/llama-index-embeddings-fireworks (#11296)

* use recursive download to get all files

* fix constants

* remove branch and repo constants

* remove f string

Update simpledirectoryreader.md

cr

* Init colbert reranker

* cr

* cr

* cr

* fix

* cr

* cr

* cr

* cr

* cr

* cr

Fix: TypeError: chromadb.api.client.Client.get_or_create_collection() argument after ** must be a mapping, not NoneType (#11104)

* update

* address comments

Bump cryptography from 42.0.2 to 42.0.4 in /llama-index-integrations/agent/llama-index-agent-openai-legacy (#11105)