fix: /tmp/hardhat-solc permissions to be world writable

[KartikSoneji]

• Because this PR includes a bug fix, relevant tests have been included.

---

Fixes: #7161

Use chmod() to set permissions to 777 on the /tmp/hardhat-solc folder, if possible, to ensure it's accessible.

[alcuadrado]

Hey, thanks for this PR. I'll take a look soon.

Another alternative would be to stop using tmp dirs and use <project>/cache/tmp or something like that. So that we avoid these kind of problems altogether. WDYT?

[KartikSoneji]

Hmm the issue is it would force everyone to update their .gitignore files.

[alcuadrado]

but everyone has cache/ in their gitignore files, we add it during initialization, and hardhat would be really annoying without it

[KartikSoneji]

Oh ok.
Then that might be a better solution.
Should I make the changes?

[alcuadrado]

if you want, that'd be cool.

I think in our own tests it's still fin to use tmp dirs. But in the actual code we should probably have a helper that creates them inside the cache dir. We have this:

export async function mkdtemp(prefix: string): Promise<string>;

Maybe it could be

export async function mkdtemp(cointainer: string, nameHint: string): Promise<string> 

so that we call it with hre.config.paths.cache as container, and it creates a new folder base on the nameHint.

[KartikSoneji]

Should it be tmpDir instead?

So instead of:

export async function mkdtemp(prefix: string): Promise<string> {
  try {
    return await getRealPath(
      await fsPromises.mkdtemp(path.join(tmpdir(), prefix)),
    );
  } catch (e) {
    ensureNodeErrnoExceptionError(e);
    throw new FileSystemAccessError(e.message, e);
  }
}

It can be:

export async function mkdtemp(
  prefix: string,
  tmpDir: string = tmpdir(),
): Promise<string> {
  try {
    return await getRealPath(
      await fsPromises.mkdtemp(path.join(tmpDir, prefix)),
    );
  } catch (e) {
    ensureNodeErrnoExceptionError(e);
    throw new FileSystemAccessError(e.message, e);
  }
}

[kanej]

Hey @KartikSoneji, thanks for all the help here.
I am closing this PR as we tackled #7161 with this PR: #7220