Refactor _process_results to support new DB engines

[dantownsend]

I'm one of the developers behind psqlpy. We've created a third-party library, which you can check out here. However, we've encountered a small problem that makes the integration nearly impossible without your assistance.

The issue lies within the _process_results method, where some unwrapping is performed. Asyncpg returns a Record object, but psqlpy does not. Instead, it behaves more like SQLite in this context, returning a list of dictionaries. Therefore, we don't need this unwrapping. Unfortunately, there is no appropriate engine type for our use case. We cannot use the SQLite engine type for obvious reasons.

Would it be possible to add a psqlpy-postgres type, so we can eliminate this Record object unwrapping?

Additionally, this change needs to be reflected in the Query class within the querystrings property. There might be other places affected, but these are what I identified during my reverse engineering.

Looking forward to our collaboration!

Originally posted by @insani7y in #986 (comment)

As outlined above, there's a bit of code where we transform the response from the database into a list of dicts:

piccolo/piccolo/query/base.py

Lines 51 to 56 in e7b13fb

	if self.engine_type in ("postgres", "cockroach"):
	# asyncpg returns a special Record object. We can pass it
	# directly into zip without calling `values` on it. This can
	# save us hundreds of microseconds, depending on the number of
	# results.
	raw = [dict(zip(keys, i)) for i in results]

It assumes that Postgres engines will just be asyncpg based, which is no longer the case.

We should move this logic to the Engine class.