Some small stuff, including the revert of the seqan3::record you suggested.
This also add the concept checks to var_io::reader_options which is similar to map_io in complexity. Compile time for var_io_reader_test increases from 11.19s to 11.47s (median out of three each). And this is for a test that actually instantiates multiple different options whereas users usually only instantiate options once. So I think this is very negligible.
Some small stuff, including the revert of the seqan3::record you suggested.
This also add the concept checks to var_io::reader_options which is similar to map_io in complexity. Compile time for
var_io_reader_test
increases from 11.19s to 11.47s (median out of three each). And this is for a test that actually instantiates multiple different options whereas users usually only instantiate options once. So I think this is very negligible.