Prerelease String shouldn't always be treated as lower version

Hey,
I've found some irregularity where I have the following two versions for tcpdump
`4.9.3-4ubuntu0.1`
`4.9.3`

Using the example code from the readme we get that `4.9.3` is greater than `4.9.3-4ubuntu0.1` even though that's not the case
[Link to playground](https://go.dev/play/p/lYLfj29avfs) to see my tests.

I'm pretty sure that `4ubuntu0.1` doesn't mean it's a prerelease version and it's a widely used suffix.
In other compiled binaries for ubuntu.

I'm not sure what's the correct approach to solve this, but I was wondering whether we should narrow down the behavior when evaluating the prerelease string.
WDYT?

Start with the README example and the linked Go playground to reproduce the comparison between 4.9.3-4ubuntu0.1 and 4.9.3. Determine the intended handling of the Ubuntu suffix, then add coverage for the reported ordering and confirm the version comparison behaves accordingly.

Written by the indexing model from the issue text.