v.db.select mixes areas and other geometries resulting in a wrong region

[marisn]

v.db.select with option "print region" calls Vect_cidx_find_all with type mask -1. This type mask matches lines and areas (see #3235). As a result - output of it will be IDs of lines and IDs of areas. Subsequent calls to Vect_get_line_type and Vect_get_line_box assume that Vect_cidx_find_all has returned only line IDs and thus will read wrong line for matching areas. The solution would be to change type mask to read "~GV_AREA" instead of "-1".

I have not commited this as I would like if someone with greater understanding of CIDX to take a look at it at first (together with #3235). Here is sample data to test the issue. In both cases output should be identical, still, as you can see, for "bad" map it is wrong.

echo "ORGANIZATION: Maris N
DIGIT DATE:   01/01/2017
DIGIT NAME:   -
MAP NAME:     fail
MAP DATE:     2017
MAP SCALE:    10000
OTHER INFO:   Test polygons
ZONE:  0
MAP THRESH:   0.500000
VERTI:
B 4
 15 10
 10 10
 10 15
 15 15
C 1 1
 12 12
 1 3
C 1 1
 17 12
 1 4
B 4
 15 10
 20 10
 20 15
 15 15
B 2
 15 10
 15 15
" | v.in.ascii in=- format=standard output=bad --o
v.db.addtable map=bad

echo "ORGANIZATION: Maris N
DIGIT DATE:   01/01/2017
DIGIT NAME:   -
MAP NAME:     success
MAP DATE:     2017
MAP SCALE:    10000
OTHER INFO:   Test polygons
ZONE:  0
MAP THRESH:   0.500000
VERTI:
B 4
 15 10
 10 10
 10 15
 15 15
B 4
 15 10
 20 10
 20 15
 15 15
B 2
 15 10
 15 15
C 1 1
 12 12
 1 3
C 1 1
 17 12
 1 4
" | v.in.ascii in=- format=standard output=good --o
v.db.addtable map=good

v.db.select -r map=good where=CAT=4
v.db.select -r map=bad where=CAT=4

[mmetz]

In 70406:

v.db.select: fix #3236

[mmetz]

In 70407:

v.db.select: fix #3236 (backport from trunk r70406)

[mmetz]

In 70408:

v.db.select: fix #3236 (backport from trunk r70406)