The following:

"Prior to first access, the content property must be initialized in a way that is equivalent to processing these steps:

Input
TEMPLATE, the template
Output
CONTENTS, the template contents
Let DOCUMENT be a new Document node that does not have a browsing context
Mark DOCUMENT as an HTML document
Let CONTENTS be a new DocumentFragment object whose node document is DOCUMENT"

Implies 

template.contents.ownerDocument !== template2.contents.ownerDocument

for any two distinct template elements.

This seems needlessly expensive. I propose something equivalent to:

-Event document which *has a browsing context* owns another (lazily created) "inert document" which lacks a browsing context. When a template element creates it's content document fragment it assigns it's node (owner) document as follows:

If template.ownerDocument has a browsing context, use that document's "inert document"
else use the template.ownerDocument.