lazy_static 1.2 bumped its MSRV to Rust 1.24.1, so we rollback to using
lazy_static 1.1 when testing on Rust versions older than 1.24.1.

Add the Unicode license to the unicode-tables directory, as the data
there comes from the Unicode Consortium.

Fixes #530

This commit adds several emoji properties such as Emoji and
Extended_Pictographic. We also add support for the Grapheme_Cluster_Break,
Word_Break and Sentence_Break enumeration properties.

Following the style to the nearby links to LICENSE-APACHE
and LICENSE_MIT, the link to LICENSE-UNICODE was placed
after an opening parenthesis, but the closing parenthesis was
missing.

See also #535

PR #536

PR #515

PR #531

It is useful to access the pattern provided to the regex set.

PR #529

Shaves of a few percent on dfa heavy regexes such as `RegexSet` use.

Avoids frequent re-allocations while the instructions are pushed to it.

This was motivated in part for using `From<&[T]> for Arc<T>`, which
technically only requires 1.21.0. However, lazy_static 1.2, which is a
dependency of regex, now requires 1.24.1, so we might as well ride the
wave.

rustfmt is not something I'm willing to adopt at this time, but
contributors are still trying to use it anyway. This is irksome, so we
attempt to prevent this from happening by creating a config that
explicitly disables rustfmt.

We also add a few missing entries.