int32 in place of double to reduce storage #1488
Description
As discussed in #1486 (comment), it looks like that we could reduce storage significantly (i.e. by a half) by saving coordinates in int32 in place of double, which does not seem to lose any precision.