You can clone with
HTTPS or Subversion.
I have https://github.com/vim-scripts/ShowMarks.git as a git submodule in ~/.vim/bundle/ShowMarks and notice that whenever I try to invoke the help section for a topic, vim spins at quite too many CPU cycles to be healthy i.e. vim is unable to process anything else until a CTRL-C is used. I have a few other plugins installed similarly, but it seems that ShowMarks is the only one giving me trouble so far.
I'm not sure which module causes the trouble in this instance as I don't see a problem when ShowMarks is installed from http://www.vim.org/scripts/download_script.php?src_id=3342 to the regular ~/.vim/plugin/ directory but then again, no help file is distributed with it here (it's the git version that has one). I also don't see a problem when I comment out the call to pathogen#helptags() in my .vimrc.
https://github.com/shalomb/.vim/blob/master/messages is my output from vim when :set verbose=9 and :debug help joinspaces. I don't see anything out of the ordinary in the output, apart from what seems to be multiple attempts to locate the tags files for each module. Is this normal?
Looks like I tried to reproduce this months ago and failed, but forgot to post back to the issue. I just made a fresh attempt to reproduce it and I still can't. Note the Git version isn't distributed with a help file either; instead, it does some weird self mutilation to generate one (check git status in the repository after invoking Vim), and I'm guessing that's somehow to blame. You can try to narrow it down by killing the showmarks tags file and seeing if the issue persists. Other than that I'm tempted to pin the blame on showmarks and move on.
@shalomb, I raised the self-modification issue in the scraper repository. The suggestion was to fork the repo and check the files in after the expansion. I never got around to doing this because I stopped using submodules and it was no longer an issue for me. However some of my workarounds might help you.
I tried to contact the developers but their email addresses were no longer valid. I sent them a vimball, that they could put up in vim.org so that it would be correctly processed by the scrapper. If you want, you can give a try to that, You'd need to do this:
python unball.py showmarks.vba
At least this way we could rule out the peculiar installation mechanism of this script as the cause of your problem.
Actually, @harleypig forked the repository and fixed the layout, so I guess the simple test is to try his fork instead.
I don't know that what I did would fix this problem ... I haven't experienced a high cpu load with vim in any way. All I did was separate the documentation from the code so the module would work with pathogen. But I'd be curious to know if it did help.