3.3.3 - New feature, bug fix and text fix.

Added the InSAR Report module: generates a PSI summary sheet for a
study area (hand-drawn on the map, loaded from a polygon file, or
picked from a layer already in the project), with PS count, density
(PS/kmq), areal coverage (buffer method, configurable radius,
automatically computed in a metric/projected CRS even if source data
is in geographic coordinates), kinematic coherence (configurable
threshold), mean visibility ± standard deviation (from a VIS layer,
filtered to the study area) and mean velocity ± standard deviation
(total and coherent-only) for both ascending and descending datasets,
plus the mean EWUD vector (angle ± circular standard deviation,
velocity ± standard deviation, E and U components ± standard
deviation, all filtered to the study area even if the source EWUD/VIS
layers cover a wider extent). The EWUD angle uses circular statistics
(vector mean) instead of an arithmetic mean, correcting a bias near
the 0°/360° boundary; the resultant length R (directional coherence
indicator) is shown as a tooltip. No new dependencies required. Same
feature as InSAR Suite 3.2.7 for QGIS 3.

Added acceleration detection to the InSAR Report module: for each
dataset (ascending/descending), a piecewise linear regression with
exactly 2 fixed segments (1 breakpoint, whose optimal position is
still found automatically) is fitted to the mean coherent-PS time
series, after removing the seasonal component (thermal/hydrological,
typical of PSI data; period estimated from the actual acquisition
cadence). The ratio between the two segments' velocities is reported
(configurable threshold, default 1.5), along with the breakpoint date;
a direction reversal between the two segments is flagged separately.
The segment count is deliberately fixed at 2, not BIC-selected: a
meaningful velocity ratio requires exactly two velocities to compare,
and free-knot piecewise models tend to make BIC favor more segments
than genuinely warranted (see below). Also added the number of EWUD
cells falling within the study area, shown next to the mean EWUD
vector.

Bug fix: corrected a wrong pwlf attribute name (".rss" instead of
".ssr") in the TS "Analisi non lineare piecewise" tool, which caused
its BIC-based automatic segment count selection to silently fail
(always falling back to the same default instead of genuinely testing
segment counts via BIC). Also fixed non-deterministic results from
pwlf's optimizer by setting a fixed random seed, and removed the
seasonal component before the fit for the same reason as above. The
default for "Numero massimo di segmenti" is now 2 (was 5), still
adjustable by the user; note that, unlike the Report module, this tool
still uses BIC to pick among 2-5 segments and can occasionally favor
more segments than a geologist would expect, due to the free-knot BIC
effect described above -- this is a known limitation of the underlying
method, not a bug, and is left to the user's judgement for now. Also
fixed three unqualified Qt6 enums in the InSAR Report module
(QgsUnitTypes.AreaSquareMeters, Qt.Key_Escape) missed by the earlier
QGIS 4 porting pass, flagged by the plugin repository's Qt6
compatibility check.

Text fix: corrected inconsistent terminology used in the InSAR Report
module's interface and documentation, for consistency with the rest of
the plugin. No functional change.

None