[BastiMila]

Klipper/AFC: send the spool's weight, temperatures and brand - still nothing to install

Follows up on #31 and the Klipper backend in 1.73.0. Builds on its object-list
detection and adds what the old TT_SET_LANE macro used to carry, using only
commands AFC and Klipper already have.

I know weight sharing is not in your main fokus

What an AFC lane now receives

After SET_MATERIAL / SET_COLOR, in the same printer.gcode.script:

SET_WEIGHT LANE=<lane> WEIGHT=<g>
AFC_SET_SPOOL_TEMP LANE=<lane> EXTRUDER_TEMP=<t> BED_TEMP=<t>
SAVE_VARIABLE VARIABLE=tt_<lane>_brand VALUE="'<brand>'"    # only with [save_variables]
SAVE_VARIABLE VARIABLE=tt_<lane>_total VALUE=<full g>       # only with [save_variables]

• Weight comes from a source the user picks under Settings > Account,
  "Weight from: Chip / Cloud" (NVS wcloud, default Cloud). Cloud reads the
  spool's document users/{uid}/inventory/{chip uid} (weight_available,
  measure_gr) - what a TigerScale keeps current - on its own task, started
  the moment the tag is read. Chip uses what is written on the tag and needs no
  network. With no figure, no SET_WEIGHT is sent and the result screen says so
  in orange; the printer keeps its previous value.
• Temperatures are the middle of the tag's windows.
• Brand and full quantity have no field in AFC, and AFC keeps
  initial_weight at 1000 g. When printer.objects.list shows
  save_variables (Klipper's own section, present on OpenCreator), they go
  there and are read back. Without it they are simply not sent.
• Happy Hare is unchanged: MMU_GATE_MAP takes no weight, and no warning is
  shown for it.

On the device

• Slot grid: each colour block shows the grams left and is filled from the
  bottom to the percentage. Both are written into the existing widgets, so the
  grid is not rebuilt while a print consumes filament. The brand is shown under
  the slot again.
• Result screen: the weight that was sent is added to the line under the spool
  ("Longer · 1.75 · 500 g").
• Reader screen: the remaining weight from the chosen source.
• If the first account read brings no figure (failed, could not start, or ran
  past its 8 s budget), the send asks once more and waits at most 5 s, with a
  line on the scan screen. A document that has no weight counts as an answer
  and is not asked again.
• New strings in all eleven languages; the Chinese ones use only glyphs already
  in the compiled faces.
• New previews: setacct-out, readcloud, readcloud-wait, scanweight,
  scanretry, resultweight, resultgrams.

PrinterBackend gains three defaulted virtuals (setWeightFromCloud,
weightSkipped, sentGrams) and SlotState two fields (grams,
percentKnown); no other backend changes behaviour. moonraker-ws-mock.py
understands the three new commands.

Tested

• verify.sh: build passes.
• On a FlashForge Creator 5 Pro with OpenCreator / AFC:
• Against the mock: the exact script above parsed and stored.
• Not tested: Happy Hare, a printer without [save_variables], the 5 s retry
  on a real failure.