Candidates for refactor

There are some passes in current `main` which can be moved to the `refactor` with low effort and/or upstreamed. Most of these passes missing lit tests and docs and only tested in end-to-end python pipeline.

- [ ] `PlierToScfPass` - CFG to `scf` uplifting. It doesn't depend on our python dialect (there is currently 'ArgOpLowering` python pattern here, but is should be moved to separate pass). This is also a good candidate for upstreaming
- [ ] `PlierUtil` dialect and llvm lowering. `PlierUtil` llvm is currently done as part of numba-specific llvm lowering. It should split into separate pass (it will also help with getting rid of IMEX_ENABLE_NUMBA_HOTFIX workaround).
- [ ] `ParallelToTbbPass` pass and related TBB runtime. Converts `scf.parallel` to `util.parallel` which is not strictly tied to TBB specifically.
- [ ] `UntuplePass` - expands tuples to individual values.
- [ ] `MakeSignlessPass` - Converts all integers types to signless (we have signed/unsigned integers in early pipeline)
- [ ] `MemorySSA` analysis - store to load forwarding, dead store removal, etc optimizations on memref dialect - require a lot of preparation work and RFC to upstream
- [ ] TBD