My feel is this is almost impossible to accurately model, without going full MRP like the YouGov model.
First off - just because Labour-held seats voted Leave doesn't mean most Labour voters in those seats voted Leave. Or vice versa, for that matter. In fact, looking quickly at a YouGov from June 2017, on the whole, you'd look at 65/35 Remain/Leave on average from Labour 2015 voters. And about the reverse from Tory 2015 voters.
But this would vary hugely across the country, and between constituencies. (You'd expect lots more in London to be remainy, for example)
But grab generic constituency with, say, Lab 55, Con 35, LD 5, Others 5 on a turnout of 65% which voted LEAVE 53/REMAIN 47 on a turnout of 75%. With an election turnout of 65%, this would be (out of the entire electorate): Lab 36, Con 23, LD 3, Others 3, DNV 35
With a referendum turnout of 75%, this would be (again out of the entire electorate) about 40 Leave, 35 Remain, 25 DNV
If the party voters broke as expected, you'd get 23,8,2,1 going Remain (=34); 13, 15, 1, 2 going Leave (31)
So we can guess that the non-GE-voters broke 9-1 for Leave (which fits both polling and anecdotal evidence). *
So - that's the Brexit side of things to start with. Then you have to look at the Remain/Leave voters and ask:
- Which Labour Leavers would ever contemplate voting Tory? Worst case, they stay at home.
- Which Tory Remainers would ever contemplate voting Labour? Worst case, they stay at home.
- How many Labour voters put Brexit high enough on their internal preference sheet that they'd override the rest of the prospectus (eg "I've been waiting for a true left-wing Labour leader for decades now, I love Corbyn, the tax and nationalisation agenda speak to something deep in me... but fuck it, I'll vote for May and the Tories because Brexit")?**
- How many Tory voters have Brexit high enough on their internal preference sheet that they'd override the rest of the prospectus (e.g. "I'm totally terrified by Corbyn and McDonnell, they'd devastate me and the economy, fucking hell Venezuala, but... well, Brexit, right, so I'll vote for them anyway")?**
- How are they distributed between constituencies?
My instinct is you're looking at net shifts of maybe single figure number of constituencies right on the edge, but that's sheer gut instinct and very possibly subject to my internal biases.
* Simplified, because some voters in the GE won't have voted in the Referendum, for example, but the overall thrust is about right
** The concept that Brexit doesn't override all other considerations is anathema on the Lib Dems Newbies group, but I can be dirty and admit it here...