Python bindings: add gdal.Run() to run GDAL algorithms

[rouault]

Fixes #11832

CC @dbaston

[coveralls]

coverage: 70.691% (+0.004%) from 70.687%
when pulling b8c483b on rouault:fix_11832
into f401ea8 on OSGeo:master.

[dbaston]

I see the "run" function doing three separate things:

• providing a simplified interface to call an algorithm
• providing a way to collect all of an algorithm's outputs
• providing a means to automatically clean up any output datasets (via Finalize)

A user may want to do any combination of these, so I would look at separating them. Maybe something like:

gdal.run -> constructs an algorithm, runs it, and returns it
with gdal.run -> same as above, but calls Finalize at end of block
alg.outputs(parse_json=True) -> returns a (possibly empty) dict of an algorithm's outputs, exception if alg has not run yet
alg.output(parse_json=True) -> returns the single output or exception if count != 1

[rouault]

Thanks for all the suggestions @dbaston. They are now implemented

[rouault]

I've also added the capability of doing gdal.Run("gdal", "info", input="in.tif"), that is use *args trickery to collect unnamed arguments and assume they are the path. I find that slightly more user friendly