- fixed incorrect width of unused columns between data flag values (2 instead of 1). This allows the second prediction flag to be read in.
- Consolidated definitions of columns for maintainability.
- Updated column/index definitions to clearly define a FLAGS index, and specify names of individual flag columns.
- Fixed SP3 output generator to access and output FLAGS data.
- Addressed misalignment bug caused by the combination of CLK nodata value width, and Pandas adding a space between columns. Also applied this fix to POS nodata value, as it appears the same issue would apply there.
- Added handling for missing flag values (conformance to empty strings to avoid outputting nans)
- Added placeholder TODOs for throwing exceptions if record types are encountered that we currently do not implement support for (EP, EV).

…avoid DataFrame.to_string() inserting spaces.

…d has same effect as default.

…ed in pull #38.

…ppy. (Note that it seems to still work without this)

…e caused a crash)