- Switch TypeId to 128 bits
- Hack around the fact that tracing-subscriber dislikes how TypeId is hashed
- Remove lowering of type_id128 from rustc_codegen_llvm
- Remove unnecessary `type_id128` intrinsic (just change return type of `type_id`)
- Only hash the lower 64 bits of the TypeId
- Reword comment

Co-authored-by: bjorn3 <17426603+bjorn3@users.noreply.github.com>

…apkin

Use 128 bits for TypeId hash

Preliminary/Draft impl of rust-lang/compiler-team#608

Prior art (probably incomplete list)
- rust-lang#75923
- rust-lang#95845

…ock, r=cjgillot

Don't hold the active queries lock while calling `make_query`

This moves the call to `make_query` outside the parts that holds the active queries lock in `try_collect_active_jobs`. This should help removed the deadlock and borrow panic that has been observed when printing the query stack during an ICE.

cc `@SparrowLii`
r? `@cjgillot`

…Veykril

Fix rust-analyzer proc macro server

The feature now exists on `proc-macro-srv-cli`, and without it the proc macro server will bail rigth out.

CC rust-lang/rust-analyzer#14991

Do `fix_*_builtin_expr` hacks on the writeback results

During writeback, we do `fix_{scalar,index}_builtin_expr` so that during MIR build we generate built-in MIR instructions instead of method calls for certain built-in arithmetic operations. We do this by checking the types of these built-in operations are scalar types, and remove the method def-id to essentially mark the operation as built-in and not "overloaded".

For lazy norm and the new trait solver, this is a problem, because we don't actually normalize all the types we end up seeing in the typeck results until they're copied over writeback's copy of the typeck results. To fix this, delay these fixup calls until after this normalization has been done.

This doesn't affect the old trait solver, but does simplify the code a bit IMO, since we can remove a few sets of calls to `resolve_vars_if_possible` and some `borrow_mut`s.

r? `@lcnr`