v1.5.1: one slice count, a window that stays put

A patch release with two corrections found in reading, plus real imagery in the README. No new controls; everything in the v1.5.0 manual still applies.

One slice count in every mode. The same physical axial slice used to carry two numbers: the MPR pane said AXIAL 622/801 while the Pano, TMJ, Reslice and Region scouts said AXIAL 180/801, always summing to n+1, because MPR printed the pane index and the scouts printed the voxel index. Every slice ordinal you see or save is now the pane count, 1-based, with its direction printed beside it: AXIAL 622/801 S→I (slice 1 is the most superior), coronal P→A, sagittal R→L. The letters follow the live camera, so an obliqued pane reports the direction it actually scrolls along. Composed snapshots bake the same label, so a figure never has to be re-read against the mode it came from. The slice-stack export keeps the voxel index in its filenames, as the data contract for meta.json and NIfTI, and meta.json now states the mapping. navigate_slice is unchanged (0-based pane index: the pane labelled AXIAL 622/801 S→I is index: 621); the MCP docs now say so, so an agent and a reader mean the same slice by the same number.

The window stays put. Any window you set (a preset, the sliders, the histogram lines, a mouse W/L drag) used to revert to the volume's default about half a second after every return to MPR, written as a plain number so no preset chip lit. The other modes then showed a window you never chose, which is why it was hard to pinpoint. The cause was the v1.5.0 mouse-W/L sidebar sync taking Cornerstone's own volume-attach event for a drag. It now trusts only a real drag (W/L tool active, button held on a pane). Every reset still works as before: double-click on any slider, the histogram default, Reset all, R.

Real imagery in the README. The four interface stills and the demo GIF are now shot on a real CBCT (16×16 cm, 200 µm, full head with dentition), shown with the patient's written consent and de-identified before it was ever opened; every 3D pane is the bone surface style, never a soft-tissue render. The volume itself is not in the repository and never was. The demo GIF is recorded live over the MCP bridge on that scan.

Upgrading: git pull (or download the release) and npm install, then start as usual. No data or sidecar migration; saved annotations and views are unaffected.

Research use only. Not a medical device. CBCTScope visualizes and navigates volumes. It never produces findings, measurements-as-conclusions, or diagnoses, in the UI or over MCP. Do not use it for clinical decision-making.