The problem here might be your workflow, not the new editor (although there are plenty of issues with it still, so maybe carefully consider putting it into production just yet).
When you send an email from Marketo, the actual HTML that gets sent includes all sorts of code patterns for different inboxes because email service providers just can't get along and adhere to a standard. For example, there's conditional outlook code (needlessly) wrapped around <div> b/c for whatever reason the base code for the new editor uses <div>s instead of tables (thereby creating the problem) and then wraps those <div>s in tables to solve the problem that poor code choices created in the first place.
Also, on the receiving side, each inbox is going to do it's own thing to adjust the code that is sent to fit it's particular taste. In Gmail for example, this means that it just cuts out the MSO (Outlook Conditional) code all together (which you need if you're going to open this HTML in Outlook and not have it look malformed). Each email client has it's own quirks like this so you can't really rely on downloading HTML once it hits your inbox b/c it's truly not an accurate reflection of the actual code you sent in the first place.
To solution this issue, maybe think about changing your workflow from downloading an .eml file and attaching that as a proof and send an actual email proof to each of the reviewers on your list. If you're looking to consolidate feedback, you could follow up the send with a message which CC's everyone in the review group to be able to respond to the issues in one place and keep the dialog concentrated.
In our workflow, we use Email on Acid for render testing and once we're in a good place to send the email out for review and feedback are able to create a link to share the render results (across all the tested inboxes, rather than just "this end user's") which gives the entire team a chance to review the email in the context of "all the places" instead of "just their place". We've found a lot of value in doing render testing in tools like Litmus and Email on Acid b/c you're able to review the results of an email send across multiple devices and inboxes and it's just really hard to find a pool of actual users who cover all the bases. Said another way - if you're really only getting feedback from 4 or 5 different platforms or devices, there's a ton of stuff you're already probably missing in your review cycle.
Hope this helps to make your life a little easier in the future and avoid those times of tirelessly troubleshooting something that's not really broken in the first place, but everyone is saying it looks weird and you don't know why. Let me know if you have any questions or if there's anything else I can do to help ya get this all ironed out on your end.
Adchoices