Bug: incorrect behaviour with block processor and metadata issue with splitdist

[abey79]

While building a complex command to address the painting use case (see discussion in #487), I uncovered multiple issues:

1. faulty doc management in block:

with state.temp_document(keep_layer=False) as doc:

This construct is wrong, because doc might be replaced by the nested workflow (after execute_processors). Best use state.document throughout.

2. splitdist doesn't appear to handle metadata properly

Experiment

I created a 3-color input file, with each layer named after the corresponding layer:

vpype random -l1 -n 200 -a 10cm 10cm name -l1 red color -l1 "#f00" random -l2 -n 200 -a 10cm 10cm name -l2 green color -l2 "#0f0" random -l3 -n 200 -a 10cm 10cm name -l3 blue color -l3 "#00f" write /tmp/multilayer.svg

Create 3 "well" files, using the same color names:

vpype circle 12cm 2cm 0.5cm pagesize -l 14x10cm write /tmp/well_red.svg
vpype circle 12cm 4cm 0.5cm pagesize -l 14x10cm write /tmp/well_green.svg
vpype circle 12cm 6cm 0.5cm pagesize -l 14x10cm write /tmp/well_blue.svg

Process each layer and insert corresponding well files every XXm:

vpype read /tmp/multilayer.svg forlayer eval "cur_color=_name" lmove %_lid% 1 splitdist 2m forlayer lmove %_lid% %_lid*2% read -l %_lid*2-1% /tmp/well_%cur_color%.svg end lmove all %_lid% end show

[abey79]

The splitdist metadata issue should be fixed by #501