fix: fix slice init length

[cuishuang]

The intention here should be to initialize a slice with a capacity of 4 rather than initializing the length of this slice.

[dveeden]

Can you explain the issue you're trying to fix? Is this for functionality and/or performance?

If I understand this correctly it now appends to a slice of length 4 instead of filling a slice of length 0, but with a capacity of 4? Is this a regression? if so, what commit introduced this? Maybe we could/should add tests?

Looks like 4 is probably the wrong capacity in most cases. That's enough for the header, but not for the total packet. This is better than a capacity of 0, but a slightly larger capacity might be better. Some of these are dynamic, but some of these packets have a more predictable length.

[lance6716]

Hi, please fix ci

[dveeden]

@cuishuang could you fix the failing checks?

[cuishuang]

@cuishuang could you fix the failing checks?

Sorry, it seems that my changes were incorrect. I found that my detection tool has a high false positive rate when the slice type is byte.

Sorry for wasting your time.