Fix tail calls to `#[track_caller]` functions

We want `#[track_caller]` to be semver independent, i.e. it should not be a breaking change to add or remove it. Since it changes ABI of a function (adding an additional argument) we have to be careful to preserve this property when adding tail calls.

The only way to achieve this that I can see is:
- we forbid tail calls in functions which are marked with `#[track_caller]` (already implemented)
- tail-calling a `#[track_caller]` marked function downgrades the tail-call to a normal call (or equivalently tail-calls the shim made by fn def to fn ptr cast) (this pr)

Ideally the downgrade would be performed by a MIR pass, but that requires post mono MIR opts (cc `@saethlin,` rust-lang#131650). For now I've changed code in cg_ssa to accomodate this behaviour (+ added a hack to mono collector so that the shim is actually generated)

Additionally I added a lint, although I don't think it's strictly necessary.

Alternative to rust-lang#144762 (and thus closes rust-lang#144762)
Fixes rust-lang#144755

Rollup of 22 pull requests

Successful merges:

 - #118087 (Add Ref/RefMut try_map method)
 - #122661 (Change the desugaring of `assert!` for better error output)
 - #140740 (Add `-Zindirect-branch-cs-prefix`)
 - #142640 (Implement autodiff using intrinsics)
 - #143075 (compiler: Allow `extern "interrupt" fn() -> !`)
 - #144865 (Fix tail calls to `#[track_caller]` functions)
 - #144944 (E0793: Clarify that it applies to unions as well)
 - #144947 (Fix description of unsigned `checked_exact_div`)
 - #145004 (Couple of minor cleanups)
 - #145005 (strip prefix of temporary file names when it exceeds filesystem name length limit)
 - #145012 (Tail call diagnostics to include lifetime info)
 - #145065 (resolve: Introduce `RibKind::Block`)
 - #145120 (llvm: Accept new LLVM lifetime format)
 - #145189 (Weekly `cargo update`)
 - #145235 (Minor `[const]` tweaks)
 - #145275 (fix(compiler/rustc_codegen_llvm): apply `target-cpu` attribute)
 - #145322 (Resolve the prelude import in `build_reduced_graph`)
 - #145331 (Make std use the edition 2024 prelude)
 - #145369 (Do not ICE on private type in field of unresolved struct)
 - #145378 (Add `FnContext` in parser for diagnostic)
 - #145389 ([rustdoc] Revert "rustdoc search: prefer stable items in search results")
 - #145392 (coverage: Remove intermediate data structures from mapping creation)

r? `@ghost`
`@rustbot` modify labels: rollup

Fix tail calls to `#[track_caller]` functions

We want `#[track_caller]` to be semver independent, i.e. it should not be a breaking change to add or remove it. Since it changes ABI of a function (adding an additional argument) we have to be careful to preserve this property when adding tail calls.

The only way to achieve this that I can see is:
- we forbid tail calls in functions which are marked with `#[track_caller]` (already implemented)
- tail-calling a `#[track_caller]` marked function downgrades the tail-call to a normal call (or equivalently tail-calls the shim made by fn def to fn ptr cast) (this pr)

Ideally the downgrade would be performed by a MIR pass, but that requires post mono MIR opts (cc ``@saethlin,`` rust-lang#131650). For now I've changed code in cg_ssa to accomodate this behaviour (+ added a hack to mono collector so that the shim is actually generated)

Additionally I added a lint, although I don't think it's strictly necessary.

Alternative to rust-lang#144762 (and thus closes rust-lang#144762)
Fixes rust-lang#144755

Rollup of 21 pull requests

Successful merges:

 - #118087 (Add Ref/RefMut try_map method)
 - #122661 (Change the desugaring of `assert!` for better error output)
 - #142640 (Implement autodiff using intrinsics)
 - #143075 (compiler: Allow `extern "interrupt" fn() -> !`)
 - #144865 (Fix tail calls to `#[track_caller]` functions)
 - #144944 (E0793: Clarify that it applies to unions as well)
 - #144947 (Fix description of unsigned `checked_exact_div`)
 - #145004 (Couple of minor cleanups)
 - #145005 (strip prefix of temporary file names when it exceeds filesystem name length limit)
 - #145012 (Tail call diagnostics to include lifetime info)
 - #145065 (resolve: Introduce `RibKind::Block`)
 - #145120 (llvm: Accept new LLVM lifetime format)
 - #145189 (Weekly `cargo update`)
 - #145235 (Minor `[const]` tweaks)
 - #145275 (fix(compiler/rustc_codegen_llvm): apply `target-cpu` attribute)
 - #145322 (Resolve the prelude import in `build_reduced_graph`)
 - #145331 (Make std use the edition 2024 prelude)
 - #145369 (Do not ICE on private type in field of unresolved struct)
 - #145378 (Add `FnContext` in parser for diagnostic)
 - #145389 ([rustdoc] Revert "rustdoc search: prefer stable items in search results")
 - #145392 (coverage: Remove intermediate data structures from mapping creation)

r? `@ghost`
`@rustbot` modify labels: rollup

Rollup merge of #144865 - WaffleLapkin:track-tail, r=lqd

Fix tail calls to `#[track_caller]` functions

We want `#[track_caller]` to be semver independent, i.e. it should not be a breaking change to add or remove it. Since it changes ABI of a function (adding an additional argument) we have to be careful to preserve this property when adding tail calls.

The only way to achieve this that I can see is:
- we forbid tail calls in functions which are marked with `#[track_caller]` (already implemented)
- tail-calling a `#[track_caller]` marked function downgrades the tail-call to a normal call (or equivalently tail-calls the shim made by fn def to fn ptr cast) (this pr)

Ideally the downgrade would be performed by a MIR pass, but that requires post mono MIR opts (cc ```@saethlin,``` #131650). For now I've changed code in cg_ssa to accomodate this behaviour (+ added a hack to mono collector so that the shim is actually generated)

Additionally I added a lint, although I don't think it's strictly necessary.

Alternative to #144762 (and thus closes #144762)
Fixes #144755