This is a half-hearted attempt to fix a build failure that I don't
understand in OSS-fuzz:
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=34294

cc @DavidKorczynski

I discovered these while reviewing the code to prep for the rewrite
in regex-automata.

Fixes #797, Fixes #798

Ref: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=36474
See: https://oss-fuzz-build-logs.storage.googleapis.com/log-fe51f615-a13f-4685-b8d8-de4583da1ebd.txt

This was missed in bd0a142.

Fixes #797 (again)

PR #814

This fixes the oss-fuzz build.

Specifically, the build log[1] showed this error:

    Step #3 - "compile-libfuzzer-address-x86_64": error: inherits must
    not be specified in root profile dev

So we just remove it and inline the settings.

PR #817

[1] - https://oss-fuzz-build-logs.storage.googleapis.com/log-c9b61873-8950-4a50-a729-820d5617ff7a.txt

It looks like the dead code detector got smarter. We never ended up
using the 'printer' field in these visitors, so just get rid of it.

We were previously using '/' to indicate the dual licensing
scheme, but I guess we're now supposed to use 'OR'.

PR #843

The regex compiler will happily attempt to compile '(?:){294967295}' by
compiling the empty sub-expression 294,967,295 times. Empty
sub-expressions don't use any memory in the current implementation, so
this doesn't trigger the pre-existing machinery for stopping compilation
early if the regex object gets too big. The end result is that while
compilation will eventually succeed, it takes a very long time to do so.

In this commit, we fix this problem by adding a fake amount of memory
every time we compile an empty sub-expression. It turns out we were
already tracking an additional amount of indirect heap usage via
'extra_inst_bytes' in the compiler, so we just make it look like
compiling an empty sub-expression actually adds an additional 'Inst' to
the compiled regex object.

This has the effect of causing the regex compiler to reject this sort of
regex in a reasonable amount of time by default.

Many thanks to @VTCAKAVSMoACE for reporting this, providing the valuable
test cases and continuing to test this patch as it was developed.

Fixes GHSA-m5pq-gvj9-9vr8