Parse GPX files that bind the GPX namespace to a prefix

[cpruijsen]

Fixes #242.

gpxpy/parser.py now removes the GPX 1.0 and GPX 1.1 namespaces from element tags after it builds the element tree. A document that binds http://www.topografix.com/GPX/1/1 to a prefix then parses like a document that uses it as the default namespace. Polar Access Manager writes files like this:

<ns2:gpx xmlns="http://www.garmin.com/xmlschemas/TrackPointExtension/v2"
         xmlns:ns2="http://www.topografix.com/GPX/1/1" ...>
  <ns2:trk>...</ns2:trk>
</ns2:gpx>

Before this change, a prefixed root failed with GPXException: Document must have a `gpx` root node.. Prefixed children under an unprefixed root parsed with no tracks, because find('trk') does not match {http://www.topografix.com/GPX/1/1}trk.

Elements in other namespaces (extensions) keep their qualified names, so nsmap and the to_xml() round trip do not change. notok.gpx from the issue zip now parses with its track, points, metadata and heart-rate extension. ok.gpx gives the same result as before.

test.py adds test_with_prefixed_gpx_namespace (prefixed root and children, with an extension and a to_xml() round trip) and test_with_prefixed_gpx_elements (unprefixed root, prefixed children). Both fail without the change and pass with it.

One judgement call: Topografix-namespaced elements inside <extensions> also become unqualified. They are GPX elements, and to_xml() writes them under the GPX default namespace. If you prefer to keep extension content as it is, I can limit the change to elements outside <extensions>.