Implement muxing routing cache

[JAORMX]

This allows us to have a parsed and in-memory representation of the
routing rule engine. The intent is to reduce calls to the database.

This is more expensive to maintain since we need to refresh the cache on
every operation pertaining to models, endpoints, workspaces, and muxes themselves.

Finally, this implements the interface for matchers; Currently we only have one
"catch all" matcher, but this will change in the future.

Closes #786

Signed-off-by: Juan Antonio Osorio [email protected]

[blkt]

concern: I think we should only use locking primitives from the asyncio package, but I might be wrong.
https://docs.python.org/3/library/asyncio-sync.html

[aponcedeleonch]

Because we use asyncio in other places or what advantages do you see in it? According to the docs

asyncio primitives are not thread-safe, therefore they should not be used for OS thread synchronization (use threading for that);

So I think primitives from threading are a better choice, but to be honest also not 100% sure. I can be persuaded

[blkt]

My understanding is that we're using asyncio everywhere, so we're single threaded (but still concurrent).

Using threading risks locking the only thread of execution and halt the application.

I might be wrong here, but that's my understanding of how async execution environment work in general, not only in Python.