Slightly optimized print

[WardLordRuby]

When converting the code to use multiple threads in the live stream it was needed to convert StrVec keys into an owned type, String when receiving chunks processed by threads. Thus making the latter str::from_utf8 mapping from within print, for ordering btree map keys, no longer necessary.

Also by special casing the is_empty condition of the btree map we are able to avoid placing commas depending if there is a next element in the iterator.

This provided a negligible ~0.05s run time improvement on my machine (8c/16t). Though I am not sure if there is a better way to avoid code duplication other than using format_args! that would result in more efficient assembly.