Refactor SBX binary conversion for current IO flow

[ytsimon2004]

Summary

This PR contains two fixes from ytsimon2004:fix/sbx-to-binary-v1x:

• updates suite2p.io.sbx.sbx_to_binary to match the current v1.x binary conversion handler interface used by run_s2p
• casts the scipy.stats.mode result in sparsedetect.py to an integer scalar before it is used downstream

SBX Conversion

sbx_to_binary still used the old API and called init_settings() / find_files_open_binaries(), which are no longer active in the current IO flow. That caused SBX input conversion to fail when run_s2p called the handler with (dbs, settings, files, files_chan2).

The updated SBX converter now:

• accepts (dbs, settings, reg_file, reg_file_chan2) like the other current converters
• reads input files from dbs[0]["file_list"]
• writes through the already-open binary file handles
• updates and saves per-plane db.npy / settings.npy
• computes mean images from frame sums
• counts functional-channel frames once for two-channel recordings
• keeps sbx_ndeadcols auto-detection defaulting to -1
• keeps sbx_ndeadrows defaulting to 0, so row auto-detection only happens when users explicitly set db["sbx_ndeadrows"] = -1

Sparse Detection

estimate_spatial_scale() returned the mode(..., keepdims=True) output as a NumPy array. Later code expected an integer scalar, which could raise:

TypeError: only 0-dimensional arrays can be converted to Python scalars

The fix returns int(im.item()).

Validation

Validated by successfully running Suite2p on .sbx files from both single-plane and multi-plane recordings