fix(job-status): current layer stuck at 0 during print

[gmmcosta15]

Description

• Feature
• Bug fix
• Code refactor
• Documentation

Current layer showed 0/total for the whole print. Regression from #280: _reset_job_display() cleared layer_fallback, and the Z-estimate could only be re-armed by a print_stats.info update, which Klipper never sends again when the slicer omits SET_PRINT_STATS_INFO (unchanged field, so Moonraker never re-emits it).

• Dropped the layer_fallback gate for _reported_layer (sticky, None when Klipper reports nothing).
• _refresh_layer_display() mirrors Mainsail's pure getters: info.current_layer wins, else the Z estimate, recomputed on every filament advance and on metadata arrival.
• _max_layers() follows getPrintMaxLayers precedence: info.total_layer -> metadata layer_count -> geometry estimate.
• _layer_from_z() keeps the last displayed layer when inputs are missing, so the counter never flickers back to 0.
• Guarded on_fileinfo() by filename: fileinfo is a global signal, so browsing a file mid-print was overwriting the running job's totals.

Motivation

The counter must be stable and match Mainsail, which computes it as a derived value every frame rather than behind a one-shot arming flag. Any flag armed only by an info delta is unreachable on printers whose gcode has no SET_PRINT_STATS_INFO, which is the field case that broke.