Enable detection of bmaptool, and if present generate sparse files.

[massar]

This uses https://github.com/intel/bmap-tools to generate a .bmap file
which can be copied around. The bmaptool allows storing sparse files
onto for instance an SD card, thus avoiding having to transfer the full
disk image every time one burns one.
Thus saving lots of time (and SD card cycles).

This also adds some instructions for then transfering and copying the
sparse file.

bmaptool: info: discovered bmap file '/Users/jeroen/berry.img.bmap'
bmaptool: info: block map format version 2.0
bmaptool: info: 7324219 blocks of size 4096 (27.9 GiB), mapped 1539940 blocks (5.9 GiB or 21.0%)
bmaptool: info: copying image 'berry.img' to file 'rdisk4' using bmap file 'berry.img.bmap'
bmaptool: info: 100% copied
bmaptool: info: synchronizing '/dev/rdisk4'
bmaptool: info: copying time: 6m 49.7s, copying speed 14.7 MiB/sec

versus copying the whole 30G which would take forever...

[massar]

mmm

{code}
(process:971): GLib-ERROR **: 10:11:49.712: getauxval () failed: No such file or directory
qemu: uncaught target signal 5 (Trace/breakpoint trap) - core dumped
{code}

and then disk full... need to change that 4G to something bigger it seems ;)
Updating travis... to do 8G (the sparse files should help)

Note that it needs more disk space anyway due to tensorflow models...