seq: errors on infinitesimal scientific notation argument

[jfinkels]

Steps to reproduce:

$ seq 1e-9223372036854775808

(9223372036854775808 is 2^64)

What happens now: uutils seq terminates with an error message:

./target/debug/seq: invalid floating point argument: '1e-9223372036854775808'
Try './target/debug/seq --help' for more information.

What I expected to happen: GNU seq terminates successfully with no output (because the argument is less than 1).

Notes: this is causing a failure in GNU test file tests/seq/seq-precision.sh.