Float types and other fixes

[umlaeute]

• fixes type of signal holder (Closes: https://github.com/ericlyon/pd-lyonpotpourri/issues/26)
• generally use t_float instead of float
• use garray_resize_long() instead of the old and deprecated garray_resize() (the replacement function has been around since Pd-0.43)
• use #include rather than the non-standard #import
• initialize some variables that might otherwise go uninitialized