docs: fold field feedback into the Polar Alignment guide

[brickbots]

Summary

Early users of the Polar Alignment tool reported back with the technique that actually works in the field, plus the failure mode that most often ruins a session. This PR folds that feedback into the user guide's Polar Alignment section.

• New "Getting a good measurement" sub-section with per-mount advice:
  • EQ mount: point the Dec axis ~7–10° off the mount's polar axis (wherever it points now), sweep 30–45° in RA with the middle capture screen-vertical and the ends 15–22° either side, and use Roll Off.
  • Platform: aim within ~5° of Polaris, keep Roll On, and set the screen vertical at mid-travel so the alt/az arrows match real directions.
  • Realistic accuracy expectations: 20–30 arcmin typical, down to ~10 with a very rigid PiFinder-to-scope connection — not a polar-scope replacement, but plenty when the pole is hidden or to get Polaris into a polar scope's reticle.
• New note at the adjustment stage: lock both mount axes before touching the adjusters — a slipping RA clutch (unlocked moments earlier to rotate between captures) was the most common field failure — and on a platform, adjust only the platform, never the scope.
• Prerequisites: turn off sidereal tracking before starting.
• Reversed one piece of existing advice: step 1 used to say to aim well away from the pole. Field experience contradicts this — flexure (felt mostly in roll) grows with how much the PiFinder's attitude changes between captures, so moderate near-pole sweeps beat wide ones. Step 1 now points at the new tips sub-section.

Not included (possible follow-up)

The same discussion suggested putting a short version of the lock-your-clutches / tracking-off reminder on the ADJUST screen itself, varying by EQ/Alt-Az mode. That's a code change to ui/polar_align.py, left out of this docs-only PR.

Testing

• sphinx-build -b html -n (nitpicky mode) runs clean — no warnings, both new :ref: targets resolve.

🤖 Generated with Claude Code