refactor: address b3 propagation TODOs

[eduardoemery]

The TODO left was:

TODO: Review this logic, maybe make it more robust. Question: are this logic valid if any of the getHeader operations returns a string[].

I checked and string[] has the same behavior in this logic. getHeader() returns a string | string[], the verification isNaN(Number()) works for both.

[kjin]

Thanks for the explanation in the PR abstract. Does this mean that if we have multiple headers under the X_B3_SAMPLED key, that none of them will end up being used as the options value? Would it make more sense to always use the first value?

[eduardoemery]

Added a few lines to make sure we get the first header if it comes as an Array.