* upstream/main: (50 commits)
  ENH: Improve OPM auditory dataset and example (mne-tools#12539)
  MAINT: Bump to latest pydata-sphinx-theme (mne-tools#12228)
  MRG: Simplify manual installation instructions a little by dropping explicit mention of (lib)mamba (mne-tools#12362)
  fix PSD weights handling when bad annotations present (mne-tools#12538)
  Fix phase loading (mne-tools#12537)
  align FFT windows to good data spans in psd_array_welch (mne-tools#12536)
  explicitly disallow multitaper in presence of bad annotations (mne-tools#12535)
  MAINT: Clean up PyVista contexts (mne-tools#12533)
  MAINT: Complete API change of ordered (mne-tools#12534)
  MAINT: Reinstall statsmodels and improve logging (mne-tools#12532)
  MAINT: Remove scipy.signal.morlet2 (mne-tools#12531)
  Update README badge links (mne-tools#12529)
  BUG: Fix bug with reading his_id from snirf (mne-tools#12526)
  [pre-commit.ci] pre-commit autoupdate (mne-tools#12524)
  Fix file format check in _check_eeglab_fname function (mne-tools#12523)
  MAINT: Reenable picard in pre testing (mne-tools#12525)
  MAINT: Bump to large resource class (mne-tools#12522)
  MAINT: Restore 2 jobs on Windows (mne-tools#12520)
  Add exclude_after_unique option to mne.io.read_raw_edf/read_raw_edf to search for exclude channels after making channel names unique (mne-tools#12518)
  Improve consistency of sensor types in code and documentation (mne-tools#12509)
  ...