ignore GCC `Wdeprecated-declarations` warnings

[simonpcouch]

Closes #806 and #808, I think. Expands the existing # pragma GCC diagnostic ignored "-Wdeprecated-declarations" to much more cpp code: nearly all of nanodbc.cpp and a new section in connection.cpp.

Very much open to the idea that this may not be what we want to do long-term, though I believe will do the trick to make a release possible soon (before June 4th, so that new Snowflake wrapper can head out). Will defer to others on whether we ought to file a separate issue and get to work on addressing the use of this deprecated functionality after the upcoming release.

[hadley]

I don't think these are particularly important concerns, so I think it's fine to ignore for CRAN's sake.

[detule]

Hey @simonpcouch

On #808 I think the warnings are an indication of something perhaps a bit more nefarious - in particular that on s-u's build server the iODBC headers ( that carry the deprecation warnings ) are clobbering over the unixODBC headers. I purused this in a number of issues with Simon when I last saw it happen, and he addressed it. It looks like it has snuck in again.

[simonpcouch]

Following discussion in #806, #807, and #808, our approach here will now be:

807 - Since the MacOS builds include all debug symbols and CRAN policy doesn't permit building without them, there's not much we can do here—just note in CRAN comments. 808 - Revisit discussions with Simon re: iODBC headers if this is an issue on submission—note in CRAN comments.

That leaves 806, which is on an incoming check systems build and thus needs to be addressed. I've attempted doing so by reverting the previous deprecations changes and instead applying the existing deprecations on NANODBC_CODECVT_TYPE to the remaining three instances.

[simonpcouch]

The insight is much appreciated, @detule :)