Viewer: "Analyze dataset" starts from the defaults of a bare rugnux

The viewer's processing job had drifted from `rugnux <file>`: 60 instead of 100 first-pass rotation
images, no adaptive integration radius, a fixed spot budget instead of a measured one, the per-image
loop never capped (nthreads_auto), stills radii 4/6/13 instead of 6/8/14, ice-ring flagging taken as
off for HDF5 files that do not state it (rugnux: on for rotation), the background radial correction
off (rugnux: auto), counts_in_adu never set for marCCD/SMV, 8192 instead of 16384 FFT directions on a
CPU-only host, and a stationary goniometer axis processed as rotation.

The defaults both front ends share now live in RugnuxDefaults (spot budget, beam-stop frames,
rotation first-pass images, stills geometry frames, bare-run spot finding, stills radii, ice-ring
default, radial correction auto), and the CLI and the viewer both read them. The CLI's own behaviour
is unchanged. The viewer's only intended difference stays: it writes _process.h5.

The copied command line now leaves out -N, --max-spots and --integration-radius when the run left
those to rugnux, so it reproduces the automatic choices instead of pinning them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

viewer/JFJochProcessController.cpp

@@ -62,9 +62,11 @@ void JFJochProcessController::run_(QString file_path, DiffractionExperiment expe
         } else if (JFJochMarCCDReader::CanRead(path)) {
             marccd_reader.ReadFiles(path);
             reader_ptr = &marccd_reader;
+            config.counts_in_adu = true;   // a CCD: pixel values are ADU, as for rugnux
         } else if (JFJochSMVReader::CanRead(path)) {
             smv_reader.ReadFiles(path);
             reader_ptr = &smv_reader;
+            config.counts_in_adu = true;
         } else {
             hdf5_reader.ReadFile(path);
             reader_ptr = &hdf5_reader;