Added very first spike for NOT filtering

[alkampfergit]

This is NOT ready to be merged, because I've not updated all the Memories, but it is a possible implementation for NOT filter as for discussion #688.

We need to discuss if the interface for memory filter is ok (I've tried to maintain binary compatibility in the case someone serialized the structure somewhere). I've added a simple function to grab all the filters that are configured, you can see how you can use the helper in Redis Memory that was the only memory I've updated in this PR (if everything is good I'll proceed in implementing in other basic providers as well.

I've also modified the configuration for Redis test adding an optional setting that allows using Azure Openai for testing. Actually it is more common for organizations to have Azure OpenAI than Direct OpenAI

[alkampfergit]

@dluc I've implemented for almost all provider, except elasticsearch. The elasticsesarch provider has some test that fails outside the scope of this PR, but it uses an approac to save tag into nested object where, if we have tags with multiple values, like multiple users, it saves multiple nested object with the same name but different value. This makes quite difficult to generate the query, I'll need probably more work.

In the meanwhile you can validate the overall PR, after all elasticsearch is the only provider missing up to now.

[alkampfergit]

@dluc if you want this PR is ready for review.

[dluc]

hi @alkampfergit would you have time to merge the latest changes from main? I was trying to fix SqlServerMemory conflicts but I'm not 100% confident about my changes. Please let me know if there's anything I can do to help. This feature would be really useful, hope we can find a way to complete it.

[alkampfergit]

hi @alkampfergit would you have time to merge the latest changes from main? I was trying to fix SqlServerMemory conflicts but I'm not 100% confident about my changes. Please let me know if there's anything I can do to help. This feature would be really useful, hope we can find a way to complete it.

I'll retry in the weekend, I've tried this morning but I saw that the implementation of SQL memory changed, so I carefully need to understand what is really changed and replay my modifications.

I'll keep you informed.