Do not deduplicate diagnostics in UI tests

Error reporting infrastructure deduplicates identical diagnostics with identical spans.

While it's preferable to do this in "release"/"user-facing" mode, it sometimes brings [confusion](rust-lang#50682 (comment)) and hides details that may be important during development.

Do we run some passes multiple times when we could do it once?
How many times we run them exactly? Can this number be large? Can the multiplied error construction be expensive? Can speculative checks be made cheaper if they don't report errors?

*Relying* on this mechanism to deduplicate some specific error never looks like a proper solution to me personally.

In this PR I attempt to disable this deduplication by applying `-Z deduplicate-diagnostics=no` to UI tests.

…tboats

Relax the Sized bounds on Pin::map_unchecked(_mut)

Fixes rust-lang#67669.

…r=alexcrichton

Export public scalar statics in wasm

Fixes rust-lang#67453

I am not sure which export level statics should get when exporting them in wasm. This small change fixes the issue that I had, but this might not be the correct way to implement this.

Recognise riscv64 in compiletest

Otherwise tests can't run, fails with "Cannot determine Architecture from triple"

Unify output of "variant not found" errors

Fix rust-lang#49566.

…-Simulacrum

Build compiletest with in-tree libtest

This updates compiletest to build in `Mode::ToolStd`, using the locally-built crates for `std` and especially `test`. This way we're immune to unstable differences in the bootstrap compiler crates, whether that's a prior-release stage0 or a current release local rebuild. Fixes rust-lang#59264.

As a minor cleanup, this also removes the unused `llvm_tools` flag.