Do not reset well focus (z positions) when creating "New" protocol (if focus needs to be updated user can use "Autofocus all Steps")

[dimin]

[ericweiner]

Fix shipped -- ready for bench verification

Root cause: Protocol.from_config built labware-derived positions with z=None for every well, falling back to layer_config['focus']. Clicking New blew away whatever per-well z the user had tuned.

Fix: In-memory carry-over (Option A): Protocol.from_config reads input_config['previous_well_z']; ProtocolSettings.new_protocol builds the map via groupby Well on self._protocol.steps() before calling create_protocol. First-row-per-well wins (bottom slice for zstack, only Z for non-zstack). Explicit-positions branch unchanged.

SHA: 18e705e on 4.0.0-beta (current tip 1fde650, 4.0.0-beta14 build).

Regression test: tests/test_issue_535_z_carry_over_on_new.py (functional + AST lock on UI handler ordering).

Retest signal:

1. Open a protocol; manually focus several wells (e.g. A1=5.0, B2=7.5) -- each well's Z field reflects the tuned value.
2. Click New to create a fresh protocol on the same labware.
3. Expected: each previously-focused well retains its tuned Z (A1 still 5.0, B2 still 7.5).
4. Un-focused wells default to the layer focus as before.
5. Across-restart persistence (issue Load z-position from stored microscope.json file instead of always resetting to default value #531) is NOT covered by this fix -- in-memory only. Restarting LVP loses the tuned z values (intentional safety per the design discussion -- no auto-restore z on startup).

dimin, please download the latest beta build and verify the fix. Then either close this issue if fully resolved, or add a note describing what is still happening.

[EtalumaSupport]

Our customers do not want the previous Protocol's focus levels to be preserved at the next New. Maybe ViQi wants this but it's not good for our customers who use multiple kinds of 96 well plates with multiple sample types every day.