fix: scale exploration plot height dynamically with facet count

[Gero1999]

Issue

Closes #1283

Description

When faceting exploration plots by a variable with many levels (e.g. USUBJID), the subplots in the middle rows shrink and become unreadable. This happened because the plot container used a fixed viewport height (height = \"90vh\", fillable = TRUE) and plotlyOutput(height = \"100%\") — plotly tried to fit all panels within that fixed space.

Changes:

• Added .facet_plot_height() helper that computes a dynamic pixel height based on the number of facet panels (300px per row, minimum 500px). Returns NULL for plots with ≤4 panels so small plots are unaffected.
• Applied the helper to both individual and mean plot renderPlotly calls.
• Switched layout_sidebar containers from fillable = TRUE to fillable = FALSE and removed height = \"100%\" from plotlyOutput, so large faceted plots extend beyond the viewport and scroll naturally.

Definition of Done

• Exploration plots (individual and mean) scale their height dynamically based on the number of facet panels
• Middle-row subplots render at the same size as top/bottom rows
• Large faceted plots are scrollable within the exploration tab
• Existing plots with few facets (≤4) are not affected

How to test

1. Upload a dataset with many subjects (20+)
2. Go to Exploration > Individual Plots
3. Set Facet By to USUBJID
4. Verify all subplots render at equal size and the plot is scrollable
5. Switch to a facet variable with few levels (e.g. PCSPEC) — verify the plot looks the same as before

Contributor checklist

• Code passes lintr checks
• Code passes all unit tests
• New logic covered by unit tests
• New logic is documented
• App or package changes are reflected in NEWS
• Package version is incremented
• R script works with the new implementation (if applicable)
• Settings upload works with the new implementation (if applicable)
• If any .scss change was done, run data-raw/compile_css.R

Notes to reviewer

• The .facet_plot_height() helper uses p$facet and p$data to count panels without re-querying the data. It only activates for FacetWrap with >4 panels.
• The QC plot module (qc_plot.R) already uses a similar dynamic height pattern but computes it differently (based on subjects × groups). This helper is simpler since it only needs the facet panel count.
• Switching fillable = FALSE changes the layout behavior — the plot area now scrolls independently. Worth testing that sidebar interactions (zoom, hover) still work correctly with the scrollable container.