tinygo: add relative and absolute --dir options to wasmtime args

[ydnar]

This normalizes the --dir arguments passed to wasmtime for both wasip1 and wasip2 when running tinygo test or tinygo run. The additional --dir and PWD arguments will now be reflected when using -x.

[dgryski]

I think there is some code in the waisp1 or wasip2 branch that detects "something" (will need to double check) based on the first or last --dir option provided. Are we breaking anything else by effectively having two names for each directory in the module path here?

[ydnar]

I think there is some code in the waisp1 or wasip2 branch that detects "something" (will need to double check) based on the first or last --dir option provided. Are we breaking anything else by effectively having two names for each directory in the module path here?

It worked for the tests I ran. Did you discover anything?

[dgryski]

No, I was just wondering because I know I had issues with argument ordering. I didn't see any failures, but I can run my larger test corpus and see what falls out.

[aykevl]

@ydnar did you get any further with this PR?
You could try running the test locally (make test GOTESTPKGS=. GOTESTFLAGS="-v -run=Build"), maybe it'll give some more information as to what goes wrong?

[ydnar]

I haven't. I'll try that!

[deadprogram]

@ydnar looks like CI still failing here:
https://github.com/tinygo-org/tinygo/actions/runs/10874218316/job/30180223920?pr=4431#step:19:33

Possible timeout?

[ydnar]

Looks like a logic error on my part. Fixed!

I wanted -x to accurately reflect the arguments passed to wasmtime, but I initially failed to account for the difference between running go test and go run in the section where I moved the --dir arguments to. The last commit fixes this.

[deadprogram]

@ydnar LGTM do you want to squash these commits yourself, or would you prefer that I do it for you?

[ydnar]

Is squash merge okay?

[deadprogram]

Is squash merge okay?

Done. Thank you very much!