[release/1.6] go.mod: Bump hcsshim to v0.9.4

[dcantah]

This tag contains some fixes for hostprocess containers, mainly around
fixing task stats which regressed from a change in v0.9.3.

https://github.com/microsoft/hcsshim/releases/tag/v0.9.4

[dcantah]

close+re-open to kick off CI

[dcantah]

Linux test failures are definitely unrelated, but if anyone can re-kick them off I'd appreciate it. @thaJeztah I'm going to open up a series of commits against main to upgrade to the same tag as this PR

[thaJeztah]

#7200 (comment)

This was "semi"-intentional for now as we were looking to move main to a 0.10.0-beta tag very shortly. v0.9.2 doesn't even have the code that introduced the regression so we should be fine in that regard. If we want to do the incremental move main to 0.9.4 first as well I'm fine with that, let me know

^^ PR to main was opened and merged; #7212

[thaJeztah]

LGTM

[dcantah]

CGroupsV2 and I are in a constant feud

[fuweid]

CGroupsV2 and I are in a constant feud

Rerunning

[estesp]

hmm, the Cgroups v2 failure is potentially an interesting package version inter-dependency problem? Looks like crun is needing a specific function linkage in criu that isn't in the criu version installed? I'm not sure if this is a temporary issue with this specific mix of Fedora 35 and our scripts or something we need to fix specifically in release/1.6 (since we aren't seeing this particular failure in main)

[estesp]

Trying #7217 since the only difference now is that main has moved up to Fedora 36 as a base.

[dcantah]

Thanks Phil, it seems like CI passed for that PR

[dcantah]

Gonna rebase to bring in that fedora bump

[dcantah]

Still failed, but now it actually got past the preamble before even running the tests and failed on a port forward test

[estesp]

LGTM