adc: optimize replacement of NaNs.

When converting physical to digital sample arrays, we must replace NaN
values with the appropriate invalid-sample sentinel value.

To do this, we need to call np.isnan and use the result as a mask to
replace entries in the output array.  (Although the function
np.nan_to_num also exists, it's less efficient: it literally does just
this, but also handles infinities.)

What we don't need to do is to call any() to check whether there are
any true entries - that just means we're iterating through the same
array three times rather than once.  Furthermore, np.copyto can
broadcast d_nans across the rows of p_signal, so all the channels can
be handled at once.

Also use copyto in adc_inplace_1d for consistency.

wfdb/io/_signal.py

@@ -539,7 +539,7 @@ def adc_inplace_1d(ch_p_signal, adc_gain, baseline, d_nan):
             np.multiply(ch_p_signal, adc_gain, ch_p_signal)
             np.add(ch_p_signal, baseline, ch_p_signal)
             np.round(ch_p_signal, 0, ch_p_signal)
-            ch_p_signal[ch_nanlocs] = d_nan
+            np.copyto(ch_p_signal, d_nan, where=ch_nanlocs)
             ch_d_signal = ch_p_signal.astype(intdtype, copy=False)
             return ch_d_signal
 
@@ -550,10 +550,7 @@ def adc_inplace_2d(p_signal):
             np.multiply(p_signal, self.adc_gain, p_signal)
             np.add(p_signal, self.baseline, p_signal)
             np.round(p_signal, 0, p_signal)
-            if nanlocs.any():
-                for ch in range(p_signal.shape[1]):
-                    if nanlocs[:, ch].any():
-                        p_signal[nanlocs[:, ch], ch] = d_nans[ch]
+            np.copyto(p_signal, d_nans, where=nanlocs)
             d_signal = p_signal.astype(intdtype, copy=False)
             return d_signal