Allow db2sps to use variables

[clinaresl]

db2sps currently works but it is awkward to use when designing large spreadsheets. In particular, even if the concept is simple, the specification file gets huge and involved. One among other issues is that all information is placed wrt a marker which is the cardinal point of a region. The easy concept of "going to the next line" is not implemented yet.
This suggests having "variables" which could be of two types:

• Default variables: mark (or point) and region which automatically keep track of the last writing. Hence, things like "mark + (0,2)" would mean "two lines below the last one" in spite of the specific contents written there! Another example "region.e + (2,0)" would be interpreted as two columns away to the east, again in spite of the contents written in the region.
• User variables: why not allowing the user to create her own anchors in the spreadsheet? This could then be freely used, again in spite of the specific contents of the spreadsheet.
  This would actually ease the task of designing and understanding spreadsheets. Note that the most relevant benefit of variables is that of referring to different parts of the spreadsheet without ever taking care of the specific contents written either there or here.

[clinaresl]

To make life easier, default variables such as "mark" and "region" could be also named with alias such as '.' and '*'