I'm just curious to see if this is something really worth testing, as it will require some coding to achieve.
What you ask is a cost problem in your UX evaluation.
As you know, cost = frequency (how many times a problem occurs) x gravity (the cost of the problem for you and the user).
You can have a first evaluation of the cost for the repeated field :
You can also evaluate cost for problems generated by its suppression :
If the costs seem very different, it is worth testing !
Coding the test is not really hard (with a simple js you can hide the confirmation field and copy first field value to this field).
Note : from my experience with this problem, I recommend to first optimize the "lost password" process.
I would be reluctant to do it. I assume you're trying to optimize your conversions.
There are many other things you can do to optimize conversions, but in my opinion, this isn't a pattern to break.