Normalise node type names

[knoxsp]

Remove non-alpha characters from node type names coming from the template, and make them lower caase, allowing for the template to display more user-friendly names, while maintaining compatibility in pywr

[pmslavin]

It is quite common in existing models to use numbers in node names, especially at the end to distinguish nodes within some kind of group, e.g. "Generator 1", "Generator 2", "Abstraction T1", etc. The re.sub would collapse all such names down to just the same alphabetic characters.

Is there a case for enforcing "variable name" convention instead, where digits can be present in a node name anywhere except as the initial character and underscores used to allow "spacing" between parts of a name?