Add configurable crossfade transitions between image assets

[Glassto]

Issues Fixed

Fixes #3351

Description

Adds an opt-out crossfade transition when the viewer rotates from one
image asset to the next (static image or animated GIF). Previously
this was an instant swap; now the outgoing frame fades out while the
incoming one fades in. :)

Scope is strictly image → image:

• Video and web-page assets are unaffected. Switching to/from either
  still cuts instantly, since those paths already blank the current
  image before handing back to the image loader.
• A playing GIF's own frame updates never re-trigger a fade:only an
  actual asset change does.
• The very first image shown after boot doesn't fade in from black,
  since there's nothing to fade from.

Implementation:

• image_transition.h/.cpp (new) holds the pure, unit-testable logic:
  parsing/clamping ANTHIAS_IMAGE_TRANSITION_MS (default 300 ms,
  clamped to 0–5000 ms; 0 disables it), the predicate that decides
  whether a fade should start, and the progress-from-elapsed-time
  math. This follows the same shape as the existing
  image_fallback.cpp / rotation.cpp. No QtWebEngine dependency,
  so it's covered by tests/test_image_transition.cpp (13 cases, all
  passing) without pulling in the heavier Qt modules.
• View gained the state needed to run the transition
  (fadeFromImage, a 60fps fadeTimer that only runs while a fade is
  active, fadeElapsed, fadeDurationMs, fadeActive) plus
  startImageFade()/stopImageFade().
• loadAsStaticImage() and setupAnimation() capture the outgoing
  frame right before it's overwritten and arm the fade. If a new
  image arrives mid-fade, this same capture re-seeds the fade from
  whatever is actually on screen at that instant instead of stacking
  two transitions.
• The three places that intentionally blank the image canvas
  (playVideo(), the "null" video-onset sentinel in loadImage(),
  loadPage()) now also cancel any in-progress fade.
• paintEvent() draws the outgoing and incoming frames through the
  exact same scale/centre math as before (factored into a small
  drawScaledImage() helper called twice instead of once), including
  on the linuxfb manual-rotation path, so both frames of a
  cross-fading pair are rotated identically rather than blended
  before rotation.

Not touched: asset scheduling/duration semantics (the fade happens
inside the existing display window), and the video/web-page paths
themselves.

Checklist

• I have performed a self-review of my own code.
• New and existing unit tests pass locally and on CI with my changes.
• I have done an end-to-end test for Raspberry Pi devices.
• I have tested my changes for x86 devices.
• I added a documentation for the changes I have made (when necessary).

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (master@cb28b51). Learn more about missing BASE report.

Additional details and impacted files

@@            Coverage Diff            @@
##             master    #3353   +/-   ##
=========================================
  Coverage          ?   90.77%           
=========================================
  Files             ?       85           
  Lines             ?    10475           
  Branches          ?     1163           
=========================================
  Hits              ?     9509           
  Misses            ?      707           
  Partials          ?      259           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.