@utarefson floating point string printing adds quite a bit of bloat to the library. So the option is available, but has to be added explicitly.
You can add the following to your Makefile, and the function should behave as you expected in the first example:
LDFLAGS = -u _printf_float
Alternatively, there are some helper macros for faking it without the jump in flash usage.
For example:
float foo = 0.5f;
sprintf(buffer, "Print this float: " FLT_FMT3 "\n", FLT_VAR3(foo));
(More examples of these helpers is illustrated in the seed/Logger example.)
We appreciate the feedback, and are actively working on creating better documentation inside and outside of the code.