docs: correct why the spectrum seams stay fixed and the bars vary

The note gave the reasoning for the opposite of what the code does: it
holds the seams at bar_gap and lets the bar widths absorb a pane that
does not divide evenly, not the other way round.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

plan/summary.md

@@ -1743,10 +1743,15 @@ legible. So:
   continuous, and — because the seam is a cell of the bar, not a hole — the
   full 24 bins still fit a normal pane at a pitch of two.
 - `spectrum_bar_width` became a *minimum*: bars take the spare columns of
-  their slot, seams stay exactly `spectrum_bar_gap` wide. Otherwise a pane
+  their slot, seams stay exactly `spectrum_bar_gap` wide. A pane rarely
-  that divides unevenly gives bars of different widths, and an uneven bar
+  divides evenly by the bin count — 58 columns over 24 bands gives slots of
-  reads as an uneven level — a lie about the data, where an uneven seam is
+  two and three — and the slack has to go somewhere. It goes to the bars,
-  only untidy.
+  which then vary (1 and 2 cells at that width) while the seams stay
+  uniform, because the alternative is uniform bars separated by seams of
+  one and two cells, and a seam that varies pulls the eye harder than a bar
+  that does: the seams are a repeating rhythm, the bars are data and expected
+  to differ. It is a real trade either way — an uneven bar width is a
+  slightly misleading canvas for a level.
 - The horizontal divider is `spectrum_row_gap` eighths left unlit at the top
   of every cell, so a full row draws `▇` and each value row reads as its own
   segment. One eighth costs 1/8 of a row's resolution and is the subtlest