fix(player): align edge projections with avatar drag

docs/M017_LATENCY_REPORT.md

@@ -24,6 +24,12 @@ intention. Release removes the unsent intention. Consequently a slow socket
 cannot accumulate a client-side movement backlog. No browser code predicts a
 maze position, collision result, inventory, visibility, timer, or outcome.
 
+Each received projection is translated so its authoritative `self` cell is
+directly below the fixed avatar. This replaces rectangle-centering, which could
+put the fixed avatar over a wall when a visible projection reached a maze edge.
+Drag displacement is measured from the pointer-down position rather than from
+the edge of the transparent target, preventing an unintended move on touch.
+
 ## Required deployment-network remediation and retest
 
 Do not use the M016 Wi-Fi path for latency acceptance. Place the ESP and both

docs/PLAYER_UI.md

@@ -29,6 +29,9 @@ cooldown. The browser keeps one sent move in flight and coalesces all later
 drag changes into one latest intention; release discards that unsent intention.
 The deflection is local input feedback only. The maze redraws only after a
 server-authoritative projection and the browser never predicts a cell change.
+For every projection, the browser translates only the received cells so the
+authoritative `self` cell stays under the centered avatar; this also keeps the
+avatar off a wall when the visible projection reaches a maze boundary.
 
 Commands use the version-one protocol envelope and contain intention data only: