Test records and input files

The inputs of the conformance suite and of the unit tests. Each conformance case runs in a fresh copy of this directory.

• Records written by the original C library, by tests/reference/gen_records.c: one record for each storage format (fmt8, fmt16, fmt24, fmt32, fmt61, fmt80, fmt160, fmt212, fmt310, fmt311, the FLAC formats flac8, flac16, flac24, flacmf, …), multi-frequency records (mfreq), multi-segment records of fixed and variable layout (multi, vlay, twog), skewed signals (skew), byte offsets and block-structured files (fmtblk), null signals (null), EDF files, and annotation files that use every feature of the annotation format.
• 100s: the first minute of record 100 of the MIT-BIH Arrhythmia Database, as distributed with the original package (its data/ directory), with its annotations.
• Inputs of the format converters, made by make_inputs.py (run from this directory): WAV files, EDF+ files, AHA-format files (1201.ANO, 1201.CMP), and headers in the old format.
• flacspf: 5 signals in format 508x3 (3 samples per frame), 1,200 frames, written by the original xform -M with WFDB_FLAC_BLOCK_SIZE=1001 from a record of the same layout whose samples follow a simple repeating pattern: FLAC blocks that are not a multiple of the samples per frame, on which the original library writes past the end of a buffer (see docs/KNOWN_C_BEHAVIORS.md).
• SCP-ECG files for parsescp: example.scp, a standard 12-lead record, and scp62.scp and scp63.scp, two non-standard files that parsescp rejects.

Do not change these files without regenerating the golden results of the conformance suite with the C programs (see docs/conformance.md).