omp.json: trailing buffer 4 NBSPs -> 1 (edge-to-edge powerline)

Operator 2026-05-09: "powerline doesn't go to the windows edge yet
either".

The right block had 4 NBSPs (U+00A0) of trailing buffer to push the
time segment away from WT's first-paint over-reported width.  But
the operator's `[terminal].right_margin` change to 1 (commit
2d338d1) shrunk the safety margin to 1 cell, so the powerline's
own 4-NBSP buffer is overkill -- the time segment ends up 4 cells
short of where the dashboard frame ends.

Reduced to 1 NBSP so the powerline right-edge cap aligns with the
dashboard frame's right border (both now 1 cell from the visible
edge).  Plus final_space = false (already set in commit ffc1f4f)
keeps the cursor flush after `╰─ ❯ ` without an extra trailing
space pushing the entry field column count.

If wrap returns at this width: bump back to 2 NBSPs in mios.html
(matches [terminal].right_margin which the dashboard frame uses).

Author: Kabuki94

usr/share/mios/oh-my-posh/mios.omp.json

@@ -217,7 +217,7 @@
           "//": "Trailing spacer -- 6 NBSPs (U+00A0). SSOT: mios.toml [terminal].right_margin=5 + 1 ANSI-cursor-positioning slack = 6. Total trailing buffer is 6 NBSPs + final_space (1) = 7 cells, which absorbs WT pseudo-console chrome (~3 cells: scrollbar gutter + acrylic blur + sub-pixel rounding) + invert_powerline cap over-render (1 cell) + visible-edge safety. 2026-05-09 operator screenshot at 10:02 with only 4 NBSPs: time fit cleanly at right but a 1-cell phantom blue rectangle appeared at col 0 of the second prompt line, just left of the entry field corner -- that was the right block last invert_powerline cap wrapping past the buffer to col 0 of the next row. Bumping to 6 NBSPs absorbs the cap. NBSPs not regular spaces because oh-my-posh strips trailing all-whitespace text segments before alignment math.",
           "type": "text",
           "style": "plain",
-          "template": "    "
+          "template": " "
         }
       ]
     },