[aguirreSL]

Adds the module pytta.iec61672 with the measures of a sound level meter, IEC 61672-1:2013:

1. weighting_filter(kind, fs): second-order sections of the frequency weightings A and C from the pole frequencies of Annex E (computed from E.2.4 and E.3.4, which give 20.60, 107.7, 737.9 and 12 194 Hz), through the bilinear transform and normalized to 0 dB at 1 kHz. 'Z' applies no filter.
2. time_weighted_level(signalObj, weighting='A', time='F'): exponential time weighting with the time constants of 5.8.1 (F 0.125 s, S 1 s), at every sample.
3. leq, sel, lmax and lpeak (C-weighted by default), one value per channel.
4. percentile_levels(signalObj, percents=(10, 50, 90)): the time-weighted level exceeded during N % of the signal.

Levels are in dB re the dBRef of each channel, which is 20 µPa for a channel in pascal.

The tests in tests/test_iec61672.py use the values of the standard:

1. The A and C responses at the base-ten frequencies of Table 3, from 10 Hz to 16 kHz at 48 kHz, stay within the class 1 acceptance limits. The bilinear transform lowers the response near Nyquist (about -1.2 dB at 10 kHz and -6.4 dB at 16 kHz for A at 48 kHz), which those limits allow.
2. The response to single 4 kHz tonebursts from 1 s to 2 ms follows Table 4 for time weighting F within the class 1 limits, and the sound exposure level of each burst is 10 lg(Tb / 1 s).
3. A steady 1 kHz tone of 1 Pa rms gives Leq and Lmax of 94.0 dB, an SEL over 10 s of 104.0 dB and an LCpeak of 97.0 dB.
4. 30 s at 60 dB followed by 30 s at 80 dB give L10 = 80 dB and L90 = 60 dB.

Checked with Python 3.12, numpy 1.26 and scipy 1.17. The existing tests behave as on development: the 7 failures there (3 in test_h5io.py, 4 in test_pyttasave.py) are the same with and without this change.

Co-Authored-By: Claude Fable 5.1 noreply@anthropic.com