[python-abp] Allow the parser to handle filters than begin and end with '%'

[rhowell]

Environment

Both python2 and python3, when trying to parse filters like: "%22banner%*%22idzone%", "%22request_id%*%22idzone%", and "%22zones%*%22popunder%" (used in ​https://easylist-downloads.adblockplus.org/liste_fr.txt).

How to reproduce

$ python
import abp.filters.rpy
abp.filters.rpy.line2dict("%22banner%*%22idzone%")

Observed behaviour

abp.filters.rpy.line2dict("%22banner%*%22idzone%")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "abp/filters/rpy.py", line 95, in line2dict
return strings2utf8(tuple2dict(parse_line(text, mode)))
File "abp/filters/parser.py", line 310, in parse_line
return _parse_instruction(stripped)
File "abp/filters/parser.py", line 158, in _parse_instruction
raise ParseError('Unrecognized instruction', text)
abp.filters.parser.ParseError: Unrecognized instruction in "%22banner%*%22idzone%"

Expected behaviour

Properly parse the filter:

'in': b'%22banner%*%22idzone%',
'out': {

b'type': b'Filter',
b'text': b'%22banner%*%22idzone%',
b'selector': {b'type': b'url-pattern',

b'value': b'%22banner%*%22idzone%'},

b'action': b'block',
b'options': [],

}

[abpbot]

A commit referencing this issue has landed:
​Issue 7325 - Let the parser handle filters than begin and end with '%'