Allow PyOpenWorm to be used from any directory

[pgleeson]

Expected behavior

A Python script should be able to be run from any directory and import PyOpenWorm and get access to the default database/configuration without knowing where PyOpenWorm is installed or requiring a local pow clone...

Actual behavior

Relative path to file .pow/pow.conf needs to be set in P.connect(), e.g.

https://github.com/openworm/c302/blob/907aa9de27ddbb3b3d3c1f80971f581da2db0010/c302/OpenWormReader.py#L31

Steps to reproduce the behavior

Run OpenWormReader.py above without .pow subfolder or without setting configFile

[clbarnes]

The default db/config directory, instead of being contained in ./, could be $XDG_CACHE_HOME/pyopenworm/, and whatever its windows equivalent would be. Users with a project-local cache could still specify it as such.

[mwatts15]

There are a couple of things going on here. One is that relative paths in the PyOpenWorm config are resolved against the current working directory when they should be resolved against the config file location when there is one. I can fix that in pretty short order.

For retrieving contexts from remote resources without cloning a repo first, that's TODO, but largely spec'd out, though not in Github issues yet. I think what you're calling the "default database" is effectively a cache that would be created for that use case. @clbarnes, I think that's appropriate for this use-case. Just to underscore, what's in that cache is not all of what's in a .pow directory although some files are (worm.db*, DataSource files).

[clbarnes]

It also looks like there's not much of a cross-platform standard when it comes to setting a cache directory, although I suppose so long as it's consistent for POW it's fine. If an OS manufacturer doesn't want to define a standard cache directory, then they can deal with caches in nonstandard places...

[mwatts15]

Added support in config files for referencing files relative to the config file.
cb8cb25
a8a5c03