#2 Improve general encoding management

[Manuel Collado]

The whole encoding handling policy should be
reconsidered. Al present it is not possible to process
character data outside the XMLCHARSET. This is probably
a too severe restriction.

Only UNICODE encodings are really safe for internal
processing. But this is incompatible whith simple
non-converting I/O from/to plaint textfiles or awk
source code in other encodings.

So perhaps a single encoding setting is not enough.