fix: #200 Enable use of `data()`/`size()` before trying `c_str()`/`length()`

[finger563]

Closes #200

Replace test for c_str() and length() (which only matches a few stl containers), for the more general data() and size() check.

I tested with this example application and it now works properly:

https://github.com/finger563/nimble-cpp-hid-issue-example/tree/main

Note: I realize I accidentally committed the update to NimBLEHIDDevice to expose the reportMap characteristic. Personally I think it's a good change, but LMK if you want to revert that.

[finger563]

Follow-up: if you're willing to require c++17 (c++20 is default in esp-idf v5), then we can use the concepts and requires functionality to clean up the metaprogramming.

[h2zero]

Thanks, the problem with this is that it will not work with Arduino String, which is why I used the c_str and length check since those functions are available in std::string and String.

[finger563]

@h2zero I've pushed an update which keeps support for data/size but will fall back to c_str/length and finally &v[0]/sizeof(v)

[finger563]

FYI, esp-idf 4.4 reached end of life back in July, so it is no longer supported. Might be worth upgrading your CI to allow the codebase to migrate to more modern standards :)

[h2zero]

Looks good to me, thanks!. I'll do some testing then merge shortly.

[finger563]

@h2zero Thanks, I've addressed your comments, hopefully they all pass now :)

[h2zero]

Thanks!