Panic on invalid FASTA files missing last sequence record

[bovee]

For example, a panic will be raised when parsing a FASTA like:

>1
ACGT
>2

where the last record has a newline after the header, but abruptly ends after that without a sequence.

I think (hope!) there's an agreement that this is invalid, but we should be blowing up with a ParseError instead of panicing here.