fix: Image() drops maxintval when dtype= is also given, silently truncating

[petercorke]

Summary

• Image(im, dtype=..., **other_kwargs) (eg. maxintval=, mono=, ...) applied the explicit dtype= via a raw .astype() cast before the other kwargs ever reached convert(). For an out-of-range integer downcast this truncates to the low byte instead of scaling.
• Concretely: Image(uint16_12bit_data, dtype='uint8', maxintval=4095) kept only the bottom 8 bits of each pixel, producing a corrupted, sawtooth-wrapping image instead of a properly rescaled one (convert()'s int_image() already implements the correct maxintval-aware scaling -- the constructor just never let it see the original data).
• Found via RVC3-python's visodom.py, which reads 12-bit .pgm frames with exactly this dtype=+maxintval= combination. The resulting images looked broken; idisp()/cv2.imshow's own 16-bit display conversion was independently verified correct (empirically, via on-screen pixel readback) -- the data was already wrong before it ever reached display.

Fix

When other convert()-relevant kwargs are given alongside an explicit dtype=, defer the dtype cast to convert() (which threads dtype+maxintval through int_image()/float_image() correctly) instead of astype()'ing early. The common plain-dtype-only path (no other kwargs) is untouched -- verified with a dedicated regression test.

Test plan

• New TestImageConstructorMaxintval class in tests/test_dtype_resolution.py: downscale-with-maxintval matches int_image()'s own scaling, result is monotonic (not byte-truncated), maxintval=None default still uses the source dtype's max, dtype=+maxintval=+mono= combined (the real-world case), dtype= alone stays an unscaled cast (regression guard), maxintval= without dtype= stays inert (regression guard).
• Full test suite: 999 passed, 15 skipped, no regressions.

🤖 Generated with Claude Code

[codacy-production]

Not up to standards ⛔

🔴 Issues 10 high

Alerts:
⚠ 10 issues (≤ 0 issues of at least minor severity)

Results:
10 new issues

Category	Results
Security	10 high

View in Codacy

🟢 Metrics 6 complexity · 0 duplication

Metric	Results
Complexity	6
Duplication	0

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}