Support searching and matching with perl-like regexes.

[ethercrow]

Since we already depend on text-icu.

Update 2017-09-09: We don't anymore.

[Fuuzetsu]

The regex stuff exposed by ICU is not thread-safe and has a poor interface: close to everything runs in IO. Further, I don't think there is any optimisation we can do with working on the underlying rope structure anyway for this kind of task so I don't think it makes much sense to stick this functionality in here.

[ethercrow]

Can we trust any other regex library to handle unicode properly?

[Fuuzetsu]

We only use ICU to decode/encode the file contents, in the end it all goes to Text so anything that works over Text should do fine.