Add strip_namespaces argument to Image.getxmp()

[geeno16]

Changes proposed in this pull request:

• Added an optional strip_namespaces keyword argument to Image.getxmp()
  (defaults to True, preserving current behaviour).
• When set to False, each tag keeps its full {namespace-uri}local-name
  form instead of having the namespace prefix stripped.

Currently, getxmp() always strips namespace prefixes, which can merge values from different namespaces that happen to share a local tag name. For example:

>>> im.getxmp()
{'xmpmeta': {'RDF': {'Description': {'id': ['from-a', 'from-b']}}}}

Two unrelated id tags from different namespaces (a:id and b:id) get merged into a list, with no way to tell which value came from which namespace. With strip_namespaces=False:

>>> im.getxmp(strip_namespaces=False)
{'{...}xmpmeta': {'{...}RDF': {'{...}Description': {
    '{http://example.com/ns/a/}id': 'from-a',
    '{http://example.com/ns/b/}id': 'from-b',
}}}}

[akx]

I'd only check the (constant-in-this-function) option once instead of every tag?

Suggested change

	return re.sub("^{[^}]+}", "", tag) if strip_namespaces else tag
	if strip_namespaces:
	def get_name(tag: str) -> str:
	return re.sub("^{[^}]+}", "", tag)
	else:
	def get_name(tag: str) -> str:
	return tag

[geeno16]

Done, thanks!