repair workflow and make Python code compliant with py3

[prjemian]

• FIX documentation build requires python2  #946
• FIX The workflow is failing, even when restarted. #950

[prjemian]

Now, with #946 and #950 repaired, the workflow log shows there is an indentation error in the 51 use XML entitynxdl.xsd file.

IndentationError: Something wrong with indentation on this line:
			</xs:documentation>

The code is now having troubles resolving and removing the indentation of the embedded rst documentation. Time to refactor to textwrap.dedent() from the Python standard library.

[prjemian]

Also, this code lxml.etree.tostring(docnodes[0], pretty_print=True) now returns the formatted text starting from docnodes[0] to the end of the nxdl.xsd file! That needs a fix (add method="text" kwarg) described in the lxml tutorial.

[prjemian]

@PeterC-DLS : please review

The workflow is now successful, after fixing the environment file and replacing our/my complicated (and failing) Python code that removes common indentation with similar capability from the Python standard library package textwrap.

@zjttoefs : please review
To make this workflow successful, it was necessary to resolve #946 as part of this PR. But that issue is not completely resolved since it is not possible (as demonstrated) to add a CI workflow to test with py2.7. So, I proposed in #952 to stop trying to support Py2.7 (actually, all Python versions <3.7).

[zjttoefs]

Looks good. Tried it locally and it was able to generate everything using python 3.9. Dropping support for 2.x Python makes sense.

[prjemian]

@PeterC-DLS @zjttoefs : Thanks for the quick response!