I'm replacing using the QThreadPool for handling all of the MDSplus
calls with a concurrent.futures.ThreadPoolExecutor.

There is a new function in mdsplushelpers called retrieve_all_data that
uses the ThreadPoolExecutor.  A keyword is a progress_signal. This is a
Qt signal that can be connected back to a progress bar to keep
functionality from before.  I should probably wrap the emitting in a
try except clause.

retrieve_all_data is called from an instance of a Worker using the
original QThreadPool in app.py. The done signal is connected to
handle_mdsplus_data.

I've removed handle_returning_data because it is no longer needed.  The
progress bar is updated via the new function update_progress_bar which
is connected to the worker progress signal.

I modified workers.py to pass its progress signal to retrieve_all_data.

ToDo:
    Decide on an timeout implementation detail.  I'm thinking about
    adding it to a global configuration.  It is nice to have, but it can
    take a long time to grab all of the data for long cathode plasmas.

I've decided to just leverage numpy for my type checking. I really only
want to deal with numpy arrays, so I removed the isistance check against
Iterable.

The truthyness now is simplier and just checks if len(self) is greater
than 1.  An implementation detail that I'm not sure if I like is it will
throw a ValueError if the lengths of time and data don't match.

Users can now empty the cache when they need to regrab data such as
after a change has been made on the MDSplus tree side.

The size of the cache can now be monitored via logging.

Added a function called empty_lru_cache function to empty the cache for
the decorated _retrieve_signal function.  I want the user to be able to
regrab data for the same shot if a modification has been changed on the
tree side.

Because data.Data is now truthy, I changed the check_data_dictionary to
leverage that functionality instead of checking against None.

Also, removed some old print statements.

I wanted to check the __init__ arguments to see if they are iterable. I
decorated iterable_validator with @staticmethod because it does not need
to know anything about the class name or instance.

iterable_validator will raise a ValueError if the data is not iterable.

Data class now can be 'truthy', has a defined length, and supports ==

ToDo: Redo the __repr__ function and __str__ functions.

I was encountering an error (see Issue #12 messages) where matplotlib
doesn't calculate the new subplot coordinates correctly.  There ends up
being a negative width or height followed by crashing the program.  I
don't seem to have this issue anymore by using gridspec.  However, I
needed to add some functionality to app.py to use
GridSpec.tight_layout() because Figure.tight_layout() wasn't working
like I thought would.

Created a helper function called _retrieve_signal where all inputs are
hashable so that @functools.lru_cache could be utilized.

retrieve_signal now unpacks some of the inputs and calls
_retrieve_signal.

Some of the logging messages were modified as well.