Added support for escaping within parsing formats

[ownaginatious]

It is currently impossible to parse dates which contain data matching an internal pattern used by arrow.

For example, Thursday, December 10, 2015 at 5:09pm -0700. A format string for arrow to parse this would be, MMMM D, YYYY at h:mma Z.

Unfortunately, this doesn't work because the a in at gets interpreted as the marker for am/pm parsing in accordance with the documentation here.

This pull request offers a workaround by escaping anything that is square bracketed within a format and interpreting it as only text.

This fixes the above pattern: MMMM D, YYYY [at] h:mma Z

Arrow can now successfully parse the expression:

>>> import arrow
>>> DATE_FORMAT = "MMMM D, YYYY [at] h:mma Z"
>>> my_date = "Thursday, December 10, 2015 at 5:09pm -0700" 
>>> arrow.get(my_date, DATE_FORMAT)
'<Arrow [2015-12-10T17:09:00-07:00]>'

The only caveat as that # must be escaped in all parsing formats with [#], as it's used as the replacement character when the expression is being built.