refactor respec config to avoid race condition

[deniak]

The defer attribute on the respec script can cause some race condition with spec-generator.
That PR removes the attribute and fixes the group associated with the document.
Fix w3c/spec-generator#396

/cc @kfranqueiro

[himorin]

approve, but I don't think we'll not rely on spec-generator (or echidna) but use manual publication (due to various issues, including timeout from long processing time) only once as final update, when @kidayasuo concludes discussion around #453 .
Although we cannot archive this repository, since it's the main place for JL-TF, but all contents of jlreq Note will be frozen once that publication made.

[deniak]

The fixes in that PR also apply for regular use of respec. I'm mentioning spec-generator because that's where @xfq filled out the issue.

[xfq]

defer is recommended by ReSpec: https://respec.org/docs/#including-respec-in-html-documents

Are we changing that, @deniak?

[deniak]

defer is recommended by ReSpec: https://respec.org/docs/#including-respec-in-html-documents

Are we changing that, @deniak?

I haven't debugged more but defer works well in general but in the case of this spec, the other scripts are causing a race condition.