Refactor syncuser to reduce walks of the container fs

[anderbubble]

Description of the Pull Request (PR):

This PR refactors syncuser to only walk the container fs twice: once for user changes and once for group changes. The previous implementation of syncuser walked the container file system for every user and group being sync'd. This caused syncuser to take a very long time when large images were combined with large user/group counts.

This PR also changes the (internal) semantics of syncuser to consistently use a "write" boolean throughout, rather than "write" in some places and "showOnly" in others.

To reduce some frequently-observed confusion, wwctl container import has been adjusted to only run syncuser if --syncuser is specified, rather than always running it in "showOnly" mode when --syncuser=false. This eliminates a spurious "error" report for users who do not intend to use syncuser.

This fixes or addresses the following GitHub issues:

• Closes syncuser walks the file system for each user/group #1209

Before submitting a PR, make sure you have done the following:

• Signed off on all commits (e.g., using git commit --signoff) in agreement to the DCO
• Read the documentation for contributing to Warewulf
• Added changes to the CHANGELOG if necessary
• Updated userdocs if necessary
• Based this PR against the appropriate branch (typically main)
• Added myself as a contributor to the Contributors File

[mslacken]

I don't see any comparison like if id == info.ContainerID. So every exiting file is of a uid/gid is added to db[name] during the filesystem walk?

[anderbubble]

From @mslacken in Slack:

In the syncuser code I saw following:

type syncInfo struct {
	HostID         int      `access:"r,w"`
	ContainerID    int      `access:"r,w"`
	ContainerFiles []string `access:"r,w"`
}

what does the access:"r,w" do? This looks like the #pragma oin C/C++ but I couldn't find documentation for this. Can you give me a hint?

[anderbubble]

@mslacken I looked into the access:"r,w" stuff. It looks like that originally came from 7f1636c:

7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  19) type completeUserInfo struct {
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  20)        Name        string
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  21)        UidHost     int      `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  22)        GidHost     int      `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  23)        UidCont     int      `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  24)        GidCont     int      `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  25)        FileListUid []string `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  26)        FileListGid []string `access:"r,w"`
7f1636c9a (Christian Goll 2022-03-10 12:17:13 +0100  27) }

So I don't know what it does; but it looks like you wrote it in the first place. ^_^

I removed them locally and it didn't break tests or vet. I generally avoid making unrelated changes in a PR; but I'll do so here without objection if you'd like.

[mslacken]

@anderbubble after fixing the changelog this ready to merge