Support orca's mathsymbols.py

[sthibaul]

Joanmarie would like to get rid of orca's mathsymbols.py (https://gitlab.gnome.org/GNOME/orca/-/issues/313), and it does indeed belong to speechd.

[sthibaul]

• The _alnum conversions are already handled by speechd's call to g_utf8_normalize in the server speak function.

• Some _arrows, _operators and _shapes are already handled by nvda's dictionary, others need to be imported in orca-chars.dic along with their translations. We can as well import everything and cleanup later on (because ideally we'd keep anything for which orca has a translation and nvda doesn't).

• The processing done by _combining and at the end of _getSpokenName should be implemented just before the g_utf8_normalize call, only in the message->settings.type == SPD_MSGTYPE_CHAR) case.

• Ideally we would have a math mode set by the client, that applies that processing also for the message->settings.type == SPD_MSGTYPE_TEXT) case.

[sthibaul]

I have imported _arrows, _operators and _shapes from orca 49 alpha