Fix `dtype` usage before SciPy 1.17

[tbowers7]

There are several warnings that crop up in the PypeIt unit tests about deprecated dtype usages that will need to be addressed before the release of SciPy 1.17 (around the start of 2026, based on recent release cycles -- SciPy 1.16 was released in late June 2025).

pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
DeprecationWarning: dtype=>f4 is not supported by correlate and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (wvutils.py:612)

pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
DeprecationWarning: dtype=>f4 is not supported by correlate and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (_signaltools.py:283)

pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
pypeit/tests/test_runpypeit.py::test_run_pypeit
DeprecationWarning: dtype=>f4 is not supported by choose_conv_method and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (_signaltools.py:1538)

[tbowers7]

The following show up in the Dev-Suite Unit Tests, but track back to issues in the repository:

unit_tests/test_calibrations.py::test_it_all
unit_tests/test_calibrations.py::test_reuse
unit_tests/test_ql.py::test_run_ql
unit_tests/test_runcalibonly.py::test_run_pypeit_calib_only
DeprecationWarning: dtype=>f4 is not supported by correlate and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (wvutils.py:612)

unit_tests/test_calibrations.py::test_it_all
unit_tests/test_calibrations.py::test_reuse
unit_tests/test_ql.py::test_run_ql
unit_tests/test_runcalibonly.py::test_run_pypeit_calib_only
DeprecationWarning: dtype=>f4 is not supported by correlate and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (_signaltools.py:283)

unit_tests/test_calibrations.py::test_it_all
unit_tests/test_calibrations.py::test_reuse
unit_tests/test_ql.py::test_run_ql
unit_tests/test_runcalibonly.py::test_run_pypeit_calib_only
DeprecationWarning: dtype=>f4 is not supported by choose_conv_method and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (_signaltools.py:1538)

unit_tests/test_telluric.py: 2 warnings
unit_tests/test_wmko_flux_std.py: 62259 warnings
DeprecationWarning: dtype=>f8 is not supported by correlate and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (telluric.py:337)

unit_tests/test_telluric.py: 2 warnings
unit_tests/test_wmko_flux_std.py: 62259 warnings
DeprecationWarning: dtype=>f8 is not supported by choose_conv_method and will raise an error in SciPy 1.17.0. Supported dtypes are: boolean, integer, `np.float16`,`np.float32`, `np.float64`, `np.complex64`, `np.complex128`. (_signaltools.py:1538)

[tbowers7]

Directly in PypeIt

The list of directly offending locations is:

• PypeIt/pypeit/core/wavecal/wvutils.py

  Line 612 in d414bcf

  corr = scipy.signal.correlate(y1, y2, mode='full')

• PypeIt/pypeit/core/telluric.py

  Line 337 in d414bcf

  conv_model = scipy.signal.convolve(tell_model,g,mode='same')

SciPy _signaltools.py:283

The location _signaltools.py:283 is within SciPy and is related to the correlate() function. The only other place in PypeIt that uses SciPy's correlate() is:

• PypeIt/pypeit/spectrographs/slitmask.py

  Lines 877 to 878 in d414bcf

  	xc = signal.correlate(numpy.roll(design_slits_y, (fftsize-size)//2),
  	numpy.roll(trace_slits_y, (fftsize-size)//2), mode='full', method='fft')

SciPy _signaltools.py:1538

The location _signaltools.py:1538 is within SciPy and is related to the convolve() function. The telluric.py reference above is the only location in PypeIt that uses the signal.convolve() function. The following locations within PypeIt use SciPy's ndimage.convolve() function:

• PypeIt/pypeit/utils.py

  Lines 774 to 778 in d414bcf

  	return scipy.ndimage.convolve(img, kernel, mode='nearest')
  	# Weighted smoothing
  	cimg = scipy.ndimage.convolve(img * wgt, kernel, mode='nearest')
  	wimg = scipy.ndimage.convolve(wgt, kernel, mode='nearest')

• PypeIt/pypeit/core/skysub.py

  Line 465 in d414bcf

  dsamp = scipy.ndimage.convolve(dsamp_med, kernel, mode='reflect')

• PypeIt/pypeit/core/trace.py

  Line 137 in d414bcf

  bpm_grow = ndimage.convolve(bpm.astype(float), kernel, mode='nearest') > 0.0

• PypeIt/pypeit/core/tracewave.py

  Line 770 in d414bcf

  thismask_grow = ndimage.convolve(thismask_pad.astype(float), kernel, mode='nearest') > 0.0

NOTE: There are several locations in PypeIt that use AstroPy's convolve() function, but that implementation does not use SciPy's convolve() function.