Issue 11947: re.IGNORECASE does not match literal "_" (underscore)

Regular expressions which are written match literal underscores ("_", ASCII 
ordinal 95) and specify `re.IGNORECASE` during compilation do not consistently 
match underscores: it seems some occurrences are matched, but others are not.

The following session log shows the problem:

    Python 2.6.5 (r265:79063, Apr 16 2010, 13:57:41) 
    [GCC 4.4.3] on linux2
    Type "help", "copyright", "credits" or "license" for more information.
    >>> import re
    >>> subject = "[Conclave-Mendoi]_ef_-_a_tale_of_memories_00-12_H264"
    >>> print subject.encode("base64")  # Incase my environment encoding is to blame
    W0NvbmNsYXZlLU1lbmRvaV1fZWZfLV9hX3RhbGVfb2ZfbWVtb3JpZXNfMDAtMTJfSDI2NA==

    >>> re.sub("_", "X", subject)  # No flags, does what I expect
    '[Conclave-Mendoi]XefX-XaXtaleXofXmemoriesX00-12XH264'
    >>> 
    >>> re.sub("_", "X", subject, re.IGNORECASE)  # Misses some matches
    '[Conclave-Mendoi]XefX-_a_tale_of_memories_00-12_H264'
    >>> 
    >>> re.sub("_", "X", subject, re.IGNORECASE | re.LOCALE)  # Misses fewer matches
    '[Conclave-Mendoi]XefX-XaXtaleXofXmemories_00-12_H264'
    >>> 
    >>> re.sub("_", "X", subject, re.IGNORECASE | re.LOCALE | re.UNICODE)  # Works OK
    '[Conclave-Mendoi]XefX-XaXtaleXofXmemoriesX00-12XH264'
    >>> 
    >>> re.sub("_", "X", subject, re.IGNORECASE | re.UNICODE) # Works OK
    '[Conclave-Mendoi]XefX-XaXtaleXofXmemoriesX00-12XH264'
    >>> 
    >>> type(subject)  # Don't think this is a unicode string
    <type 'str'>
    >>> 

Since my `subject` variable is of type `str` and only contains ASCII characters
I do not believe that the `re.UNICODE` flag should be required.

help(re.sub) says:

    sub(pattern, repl, string, count=0)

and re.IGNORECASE has a value of 2.

Therefore this:

    re.sub("_", "X", subject, re.IGNORECASE)

is telling it to replace at most 2 occurrences of "_".