make the default wasm execution method compiled

[coriolinus]

This promotes using wasmtime instead of wasmi for the majority of users.

Note: this PR removes the wasmtime feature entirely, and includes that dependency always.

Previous note, now obsolete: in principle it is possible for this default to result in a panic shortly after startup if a user has compiled their node without the wasmtime feature.

[bkchr]

However, that feature is enabled by default for common architectures, so this should not be a big deal in practice.

This is not correct. What you are citing there is the substrate node. The rest of the code is still not enabling the feature automatically.

I think we should remove the feature now completely and have wasmtime enabled all the time, automatically.

@pepyakin WDYT?

[pepyakin]

Can we instead just enable these features by default at the crates that declare them and do not remove them? This would allow to build nodes without wasmtime. That seems to be somewhat useful since wasmtime is a big dependency.

Not sure if that's a realistic need though. I am ok with removing the feature altogether since that would reduce the complexity.

[coriolinus]

The problem with enabling wasmtime by default is that we set default-features = false more times than not when specifying a dependency. I've got a commit building now which removes the feature entirely, enabling wasmtime permanently. Assuming all is well with the build, it should go up in a few minutes.

[coriolinus]

Looks like the wasm build is currently failing because wasmtime depends on wasmtime-cache which depends on errno which doesn't build right for that target. Working on sorting that out.

[tomaka]

The wasmtime feature exists because of the browser stuff. wasmtime doesn't compile for the browser.

[bkchr]

Alternative approach: #8855

[bkchr]

The wasmtime feature exists because of the browser stuff. wasmtime doesn't compile for the browser.

While this wasn't the reason the feature was introduced, it is now a valid reason to keep it :D

[coriolinus]

#8855 is the better way forward.