Bugs / improvements in pyfda_io_lib

[chipmuenk]

The I/O routines contain some bugs ...

• : Exporting / importing CSV coefficient files for IIR filters does not work in many cases, especially when switching between FIR and IIR filters:
  • [pyfda_io_lib.py:574] not enough values to unpack (expected 2, got 1)
  • sometimes the scale factor is wrong
  • sometimes the last cell of the a coefficients contains a ',' instead of a '.' which is misinterpreted as a complex number
• : Exporting / importing P/Z values does not remember the k factor

... and some room for improvement:

• : It would be nice to remember and suggest the last import / export format.

• : There is no warning before overwriting an existing file

• : When exporting coefficient files in various formats, some more information should be contained in the header (corner frequencies etc.). This information is compiled in the common function generate_header(). Comment characters depending on the export format are prepended (e.g. '--' for VHDL).

• : Depending on the display format selected in the input_coeff widget and the export format selected in pyfda_io_lib, coefficients are written as fractional values or in integer format. It needs to be checked whether all combinations make sense (e.g. for VHDL packages array of real needs to be genenerated instead of array of integer) or whether it would be better to constrain exports to decimal integer format.

• : The range of integers needs to be constrained

[chipmuenk]

integrate the last remaining issue into #144