I want to flag that this PR still contains the bug that I disclose here : #109

@FlorenceBerterottiere Please rebase on latest main
@antti-ethz please review and provide your assessment

Hi @FlorenceBerterottiere

I played around with it and found some unexpected behaviour - at least unexpected to me, it might be that my expectations are wrong. Anyway, reporting them all here.

(1) - Switching parameters in Center/Span mode overwrites the per-parameter scan-settings memory:

1. In Start/Stop mode, select param A: start 10, stop 20, points 5, pattern Linear.
2. Switch to param B: start 0, stop 1, points 100, pattern Scatter.
3. Switch back to A, then to B - each restores its own settings (the existing per-parameter memory, working as baseline).
4. With B selected, toggle to Center/Span, then switch the parameter to A.

Expected: A comes back with its remembered settings (10–20, 5 points, Linear), recentered on A's current value.

Actual: the card shows B's span (1), B's points (100) and B's pattern (Scatter) applied to A.

(2) - Toggling the mode rewrites the configured scan range (use a parameter whose current value differs from the range - say its value on the Parameters page is 42):

1. In Start/Stop mode set start 10, stop 20.
2. Toggle to Center/Span, then straight back to Start/Stop.

Expected: toggling the representation back and forth leaves the configured scan unchanged (start 10, stop 20).

Actual: Center seeds from the live value on entry (42 instead of 15), so after the round trip start/stop read 37/47 - the scan changed without any field being edited.

Variant: in Center/Span mode manually type Center 5, toggle to Start/Stop and back - the manually set 5 is replaced by 42.

(3) - The selected input mode doesn't survive group/namespace changes.

1. in Center/Span mode, change the Display Group (or Namespace) dropdown.

Expected: the card stays in Center/Span mode.

Actual: it silently flips back to Start/Stop.

(4) - Out-of-bounds scans are clamped silently; the executed scan differs from what the form shows, with no feedback (needs a parameter with bounds; I used "pulse time" of the Spline RF example, which has min_value=10):

1. In Center/Span mode set Center 11, Span 6 - so the requested range is 8–14, crossing the min bound.
2. Submit.

Expected: either a validation error, or at least a notice that the range was clamped.

Actual: green "Job submitted" toast, no warning. The recorded scan values run 10.0–14.0 (evenly spaced) — the executed scan is centered on 12, not the 11 the form still displays.

(5) - Minor stuff

• In Center/Span mode with Span = 0, submitting gives "Start and stop cannot be the same". A bit confusing when the visible fields are Center and Span.
• There's a minor layout issue with the Scan Parameter components now, I see a scrollbar in it and the content is not completely visible. See screenshot.