Optimized S-VHS subdeemphasis, removed VHS-HQ

[Stefan-Olt]

• I removed VHS-HQ, because it never worked correctly: The alternating luma offsets could not be detected reliable based field type or chroma phase. Using --ire0_adjust will correct luma offsets from VHS-HQ reliable and independent of any kind track/phase detection. Using the VHSHQ tape format will now result in a warning and enable --ire0_adjust. The code for the alternating luma offsets is not removed, in case other formats need it

• I optimized the S-VHS linear subdeemphasis: The previous implementation used filter tables created with LTSpice. The LTSpice simulation tables included a bandwidth limitation of the op amp that were corrected with a shelf EQ in vhs-decode, which introduced some high frequency noise.
  The new filter is based on theoretical analysis of the (previously simulated) circuit. The filter is not a "digital filter", but an "analog filter" (digital and analog refer to the type of calculation, of course this filter is completely digital). For calculating the filter for fft from the "analog" coefficients the function analog_filtfft was introduced, that uses scipy just like filtfilt to create the filter table.
  The values for these filter seem to differ not only for SP/LP, but also for different manufacturers (devices?). As newer devices have the subdeemphasis fully integrated in a IC, only for older machines these values can taken from service manuals. I included the values from a Sony (SP) and a Panasonic (SP and LP) machine, as well as hand-tuned values to a recording I made on my JVC deck

Still open:

• LP completely untested
• Tuning of the non-linear part of subdeemphasis
• Testing with recordings made on machines from different manufacturers