These tests assert what happens if a shellout command like exec throws unexpected errors by checking what is written to stderr. Progress also gets written to stderr but is not always deterministic due to the nature of calling it via setInterval. So this disables progress for these tests so the stderr assertions can be relied on to be the same.

The logging output was making bare string assertions fail

Originally this was in #7468:

We backed off of it while we were rebuilding pacote's packument cache.

Now that that's done we can assess this in isolation.  I think it makes
sense.  The packument is cached here, all this is awaiting is
normalization and ssri calculation.

The only place this potentially does anything is in the premature
loading of manifests in `#buildDepStep` when looking at problem edges.
We can just wait till we need them and not throw a ton of requests in
parallel before we actually need them.

Removing the premature loading in problem edges will have to be a
separate effort, as it is somehow load bearing