"species information" window deletes GB_INT field

[epruesse]

When overwriting a GB_INT field with an empty string from within the species information window the entry is removed from that species. This causes the window to throw an error message. Subsequent attempts to change the field cause an 'GB_INT signed 32bit out of range' error. (submitting for rc6, but it's also present for rc7)

[epruesse]

This can be reproduced with ssu-ref-98 and 100, but seems to work on smaller/older datasets.

[westram]

• scanner tries to store a pointer in an INT awar. Since [5965] writing INTs returns an error if the value is outside 32-bit-integer-range. Here the error is not even recognized, but the value is not stored.
  • write aborted ​here
  • wrong usage of INT awar ​here
• GB_write_int is planned to take int32_t instead of long, but the this was delayed to be implemented after release
• AW_awar::write_int and AW_awar::rewrite_int need to expect int32_t instead of long as well
• Special functions needed to store pointers in AWARs(!)/DBfields(?)

[westram]

• Caution: This might make it impossible (and/or dangerous) to edit any -field located beyond the 2Gb border! Bug has been present in all previous 64bit versions.

[westram]

• fixed by [6146], [6147], [6148]

[westram]

• backport by [6149]

I've tested it with SSURef_100, valgrinded, double-checked changes, but nevertheless test b4 you use it in course!