Add is_notice flag to the --classify option

[pombredanne]

I would like to have an is_notice flag to the --classify option. It is useful to more easily collect these to keep their text as a package notice text (such as with the Apache 2.0 NOTICEs)

This can be based on:

• a key, legal file with the word "notice" in their lowered file name
• a key, legal file which is mostly made of license text and mostly matched to a license notice

[tdruez]

a key, legal file with the word "notice" in their lowered file name

As a side note, the is_legal is only true when the filename "starts or ends with", see
https://github.com/nexB/scancode-toolkit/blob/c326a0b6ed7dda44e40c8a436c7005672a1a53a9/src/summarycode/classify.py#L95-L107

This would mean that a filename such as "package_NOTICE.txt" may not be flagged as is_notice with the current behavior.
Something to consider during the implementation of a new is_notice attribute.

from commoncode.resource import Resource
from summarycode.classify import check_resource_name_start_and_end
from summarycode.classify import LEGAL_STARTS_ENDS

resource = Resource(name="package_NOTICE.txt", location=".", path=".")
is_legal = check_resource_name_start_and_end(resource, LEGAL_STARTS_ENDS)
print(is_legal)  # -> False