Turns out there are operational limitations that don't allow a
deterministic removal of the 180 degree ambiguity that exists
for calculated Declination. The one previously implemented only
worked for a subset of observatories. Now, we force all calculated
Declinations to point north(ish), which  is valid for most of the
globe, and all of the USGS observatories. However, if we ever
take absolute measurements in a location where the local magnetic
field points south(ish), we can simply provide a "shift" of +/-
180 degrees.

The scale_value used to check residual observations should always
be in radians. It is a ratio, not an angle, so converting it to
degrees never made sense. Also, the SpreadsheetAbsolutesFactory
needed to NOT change the scale_value read from spreadsheets into
degrees. Finally, both the SpreadsheetAbsolutesFactory and the
SpreadsheetSummaryFactory were modified to look for spreadsheet
files more aggressively in their respective base_directories...
this was done in this commit in order to get the residual_test.py
unit tests to all pass. The only changes made to the latter were
to remove a couple of forgotten print() statements.

The calculation of scale_value in geomagio.residual.calculate() is not
quite correct, but that needs to be addressed in a separate commit and
merge request. I'm reverting the recent change here so that we don't
break the tests.

Fixes #75

When geomagio.residual.calculate() is called with adjust_reference set equal
to True, the start/end timestamps now both reflect the time of the first absolute
measurement (i.e., West Down). This is consistent with the residual absolute
spreadsheet calculations.