Bump Rust version to stable 1.90.0

[Frizi]

Pull Request Description

• Migrate to current rust stable version - 1.90.0
• Get rid of special build profile for buildscript, hopefully preventing random full-project recompiles from happening so often during development

All other code changes are caused by formatting changes, dead code removal or new clippy lint rules finding trivial replacements.

Important Notes

The few remaining #[bench] tests were the only thing that depended on running nightly. Since we don't run them in CI anyway, they are now hidden under a "nightly" flag and can be manually enabled when running them locally. We can migrate them to criterion if desired, but the scope of that change would be significantly bigger.

[4e6]

Please run the nightly workflow to verify the release artifacts

[vitvakatu]

Overnight:

• Release workflow failed with singing issues on Mac: https://github.com/enso-org/enso/actions/runs/18723268000/job/53417421813. Today’s nightly succeeded, so it might be either a temporary issue or a failure of this branch. I will run another nightly workflow to verify.
• Changelog / Standard Library Test Report failed with connectivity issue on Mac: https://github.com/enso-org/enso/runs/53426475384#r0s18
• Curious failure in JVM tests, again, on Mac: https://github.com/enso-org/enso/actions/runs/18720081176/job/53419477492?pr=14171#step:14:7653
• http-test-server didn’t start in Standard library native tests. You will be surprised, but, on Mac. https://github.com/enso-org/enso/actions/runs/18720081176/job/53419477632?pr=14171#step:14:1470

Rerunning to verify flakiness.

[kazcw]

Additionally removed wasm-bindgen-test run, since the remaining singular test was a copy of native test anyway, adding no value.

This test is important, actually. It is the crux of our current strategy for ensuring the parser cannot be crashed by inputs with deep nesting.

See here:

enso/lib/rust/parser/src/format.rs

Lines 38 to 41 in dcb9310

	/// Maximum allowed nesting depth of compound objects. This is empirically determined to be reached
	/// before stack overflow on supported targets (see [`test::test_infinite_recursion`] and
	/// [`test::wasm::test_infinite_recursion`]).
	const RECURSION_LIMIT: usize = 1024;

If we really need to remove this test, we should first ensure that RECURSION_LIMIT is well below the value that would currently allow stack overflows, to reduce the risk that future changes cause a regression in our crash-safety; and we should update the docs to no longer suggest that we are protected by this test. However lowering the limit to rely on a larger safety margin would reject more valid programs than we can accept with a level that is validated by testing.

[Frizi]

We could bring back the wasm-bindgen test, but I'd like to keep it removed just in order to simplify the CI pipeline a bit. We also run the built parser within the GUI tests, so we could create a unit test in GUI that attempts a parse of very nested expression and check for the expected "recursion limit" to be returned. Also, the stack size of wasm module is configurable.

[vitvakatu]

I think we can bring back the test and add a short manually written workflow that runs wasm-pack test for this specific file. I don’t want to bring back all the removed Rust workflow generator code.

[vitvakatu]

I’m getting a consistent failure of Nightly release workflow for this branch: https://github.com/enso-org/enso/actions/runs/18758162553/job/53520741538#step:10:6479

I have no explanation why this happens: nightly builds against develop succeed.

GitHub

Nightly Release · enso-org/enso@37e1f5a

Enso Analytics is a self-service data prep and analysis platform designed for data teams. - Nightly Release · 37e1f5a

[vitvakatu]

Exactly the same failure is now happening on develop, so I consider CI green. Let’s merge it to avoid any merge conflicts.

https://github.com/enso-org/enso/actions/runs/18767774530/job/53549155638