dmake stuff removed

In CSV import, instead of SCWARN_IMPORT_RANGE_OVERFLOW use
SCWARN_IMPORT_ROW_OVERFLOW and SCWARN_IMPORT_COLUMN_OVERFLOW that already
existed, additionally introduced SCWARN_IMPORT_CELL_OVERFLOW if single field
data exceeds STRING_MAXLEN.

Row overflow takes precedence over column overflow that takes precedence over
cell overflow.

Without this, calculation upon import may incorrectly show #VALUE!
in some cells when the iterative calculation option is turned on.

This is on par with what we do during the ods document import.

Taking a reference to double makes no sense here.  Plus the bool.

CSV and other text formats may come with line sizes >64k that so far were
truncated due to limitations in ByteString/UniString/String, even if one line
consists of several fields that each are <64k.

Introduced additional SvStream methods that read into rtl::OString and
rtl::OUString and let SvStream::ReadUniOrByteStringLine() fill solely an
rtl::OUString.

Made Calc CSV import use those.

This version scheme is used for other packages as well. It is important for
the desktop integration packages now because we started reseting buildid for
each bugfix release (fdo#47840). This value is used also for the release
number of Linux packages.

...so that tearDown does not hang waiting for (idling) process_ to terminate
if setUp throws an exception when obtaining resolver.

We already ignore textframes inside tables, the same should be done with
textboxes till they can be handled properly.

...so that the hack of including it in gtk3salnativewidgets-gtk.cxx actually works.