This draft looks good and ready for publications. 

The draft could benefit from more discussion on the maximum sizes of records and what happens when they are exceeded. 

There seems to be some opinionated "you must eat your vegetables stuff" in the draft that will just be ignored by people using this draft. For example, I understand how OSCP stapling speeds things up but I can not see a single reason why the mechanism in this draft need the deployment of OSCP stapling.