add _not yet 100% complete_ Unicode support

add _not yet 100% complete_ Unicode support

collections, statements and BLOBs seem to be working ok
though there are still some things to be done in order to make oci_bind_array_by_name() work with U-strings

Notes:
- in Unicode mode OCI8 always speaks to Oracle server using UTF-16, so all the conversions are done by the client lib.
This is why character set parameter of oci_connect() and NLS_LANG are ignored in U-mode.

- BLOBs and CLOBs behave quite differently in U-mode.
Reading data from a CLOB would result in Unicode string, while BLOBs would return binary string.
Also, all LOB utilities work with _bytes_ when BLOB is used and _characters_ when it's CLOB.
It's not that obvious, but it does make a lot of sense to me.