csv_to_wfdb ignores several parameters when writing #592

Found while reviewing #591.

csv_to_wfdb accepts several parameters that have no effect on the written record:

• dat_file_name is always overwritten with <record_name>.dat (

  wfdb-python/src/wfdb/io/convert/csv.py

  Lines 332 to 333 in b8914e1

  	dat_file_name = record_name + ".dat"
  	dat_file_name = [dat_file_name] * n_sig

  ), in both modes.
• When writing (record_only=False), only some arguments are passed to wrsamp (

  wfdb-python/src/wfdb/io/convert/csv.py

  Lines 453 to 466 in b8914e1

  	# Write the information to a record and header file
  	wrsamp(
  	record_name=record_name,
  	fs=fs,
  	units=units,
  	sig_name=sig_name,
  	p_signal=p_signal,
  	fmt=fmt,
  	adc_gain=adc_gain,
  	baseline=baseline,
  	comments=comments,
  	base_time=base_time,
  	base_date=base_date,
  	write_dir=write_dir,

  ):
  • samps_per_frame is dropped, although wrsamp accepts it.
  • counter_freq, base_counter, skew and byte_offset are dropped, and wrsamp has no matching parameters.
  • adc_res, adc_zero, init_value, checksum and block_size are computed or accepted but then recomputed by wrsamp.

One option is to build a Record in both modes and call Record.wrsamp(write_dir=...), so that writing uses the same fields as record_only=True. For parameters that don't make sense for CSV input (for example checksum or init_value, which should follow from the samples), deprecating them may be better than passing them through.