Replace almost all pp.Combine() into pp.Regex()

[embar-]

Majority of pp.Combine() are replaced into pp.Regex() for optimization.
Only this one line is not changed:

PyDBML/pydbml/definitions/enum.py

Line 56 in 2bf015c

enum_name = pp.Combine(name("schema") + '.' + name("name")) | name("name")

This will contribute to #59 solving

Some metrics:

Before	After

Before	After

[Vanderhoof]

So I took a look, there are 19 failing tests after applying this PR. But what concerns me most is that using pp.Regex negatively impacts readability of the code. After all, why use Pyparsing then, if we end up with regex's anyway. But you are right, current PyDBML performance is terrible.

But thanks a lot for the contribution! I'll have to invest some time into the problem, and will use your changes as a basis

[embar-]

Regarding better readability – you can always add comments for it.