feat: mean pooling

[triandco]

The current implementation does not include pooling. This result in wildly different embedding to fastembed python counterpart.

This pull request add a default mean pooling for token embeddings output from ort session.

This pull request includes:

• [x] Added a hard coded embedding test,
• [x] Added mean_pool function,
• [x] Applied mean_pool in embedding generation.
• [x] Implement pooling as a post processing step
• [x] Add explicit pooling option for existing models

[denwong47]

Noted, will take a look after work.

[github-actions[bot]]

:tada: This issue has been resolved in version 4.0.0 :tada:

The release is available on:

• GitHub release
• v4.0.0

Your semantic-release bot :package::rocket: