adapter: remove QGM code

[aalexandrov]

!!! EDIT: THERE IS ONE MORE PR REMOVING QGM STUFF: #17371 !!!
If this one is reverted at some point, then that one should also be included!

This also includes #17119

Motivation

• This PR refactors existing code.

Removes the experimental QGM code, as we currently don't have concrete plans to bring it to a production-ready state.

Tips for reviewer

This also includes the changes in #17119 in order to avoid rebase conflicts (that is, we should merge this after #17119).

Checklist

• This PR has adequate test coverage / QA involvement has been duly considered.
• This PR evolves an existing $T ⇔ Proto$T mapping (possibly in a backwards-incompatible way) and therefore is tagged with a T-proto label.
• If this PR will require changes to cloud orchestration, there is a
  companion cloud PR to account for those changes that is tagged with
  the release-blocker label (example).
• This PR includes the following user-facing behavior changes:
  • The experimental qgm_optimizations session variable is no longer available and the corresponding optimization stage is removed from the optimizer. For people that have not interfered with that session variable this should not cause any changes. People that have relied on the experimental outer-to-inner join simplification in the code may experience plan regressions, but this is highly unlikely and we have not provided any guarantees about this code.

[philip-stoev]

Test changes OK.

[teskje]

Thanks!

Grepping through the code, there are two instances of "qgm" left:

• The design doc, which is probably fine to keep.
• A comment in the 101 Query Compilation doc. We might want to remove that too.

  materialize/doc/developer/101-query-compilation.md

  Lines 88 to 109 in 77a6415

  	<!--
  	# Future Pipeline
  	[Diagram](https://docs.google.com/drawings/d/1Fil1-oYy3PkP3bD7WoZphW319Pj60cMAs2HcS9A21uo/edit)
  	[Design doc](https://github.com/MaterializeInc/materialize/blob/main/doc/developer/design/20210707_qgm_sql_high_level_representation.md)
  	* [`SQL ⇒ AST`](https://github.com/MaterializeInc/materialize/blob/main/src/sql-parser).
  	* Parsing the SQL query
  	* `AST ⇒ QGM`.
  	* Name resolution.
  	* `QGM ⇒ QGM`.
  	* Optimizing rewrites + decorrelation + more optimizing rewrites.
  	* `QGM ⇒ MIR`.
  	* Lowering.
  	* [`MIR ⇒ MIR`](https://github.com/MaterializeInc/materialize/blob/main/src/transform).
  	* Optimizations. What this looks like is to be determined.
  	* Some optimizations may become redundant after optimizing rewrites are added.
  	* Note that we may be able to eliminate the per-view/cross-view distinction by modifying MIR to have more than one starting point.
  	* `MIR ⇒ LIR`.
  	* `LIR ⇒ TDO`.
  	-->

[antiguru]

Thanks!