Allow wildcards for tracepoint categories

[viktormalik]

It is now possible to run e.g. this:

bpftrace -e 'tracepoint:kvm*:* { @[probe] = count(); }'

Resolves #1212.

Checklist

• Language changes are updated in docs/reference_guide.md
• User-visible and non-trivial changes updated in CHANGELOG.md

[viktormalik]

Embedded build failed on a uprobe test (cannot open mount ns), which shouldn't be affected by this PR. Not sure why the error occurred, it passed on my fork's branch build.

Also there are clang-format fails in a new test that I copied from an existing one (which contains the same format errors, too).

[fbs]

Embedded build failed on a uprobe test (cannot open mount ns), which shouldn't be affected by this PR. Not sure why the error occurred, it passed on my fork's branch build.

Some tests are still somewhat flaky and fail sporadically, rerunning it usually fixes it

Also there are clang-format fails in a new test that I copied from an existing one (which contains the same format errors, too).

We try to incrementally fix the formatting errors. So if you touch code you format it. This way we can keep git blame useful.

[danobi]

I kicked off another round of embedded tests. Let's see if it passes this time.

We try to incrementally fix the formatting errors. So if you touch code you format it. This way we can keep git blame useful.

+1

[viktormalik]

I kicked off another round of embedded tests. Let's see if it passes this time.

Thanks, they all passed now.

We try to incrementally fix the formatting errors. So if you touch code you format it. This way we can keep git blame useful.

Sounds good, ok, fixed. I'll keep this in mind.

[danobi]

This entire codepath is pretty gnarly and is in great need of a good refactor. This PR doesn't make it materially worse (in fact it might even be a little better).

LGTM but every time anyone touches this code path I get scared something's gonna break but have no real way of knowing.