This is due to the confusion around what Brexit will mean, how long it will take, how far it will go and what agreements the EU and UK will sign.
If a research partner drops out halfway through the project, the whole effort may fall apart - and with research institutions and industrial partners booking the alloted funds each month, such a failure is hard to justify for any EU officer.
Of course, it is ridiculous to simply require that the UK partners be dropped from the project without a significant change of plan or adequate substitution.
Edit: Also, the aim of these projects is always to strengthen industry or scientific progress in the participating countries - so politically, if the UK really drops out this would be giving something for free to a competitor.
Successful projects often lead to future projects, so people thinking of projects in a few years time might not want to start collaboration in the UK, at least if there's an alternative elsewhere in the EU.
¹ For example, the UK lost me — after the 2014 European Parliament elections in the UK, where UKIP won most votes, I started looking for a job elsewhere.
In general a partner dropping out of a project partway through is very disruptive, so you like to avoid it. You want partners who will still be there for at least 4 years from now: 1 year for the application preparation/review, plus 3 years for the project if the application is successful. Many people are not sure if that will be true of UK partners; can you say, today, that the UK will still be part of the EU funding mechanisms through 2020? I think there's a good chance they will be, even if they leave the EU (they will probably end up joining the research framework a la carte, like Norway, Iceland, and Switzerland have). But I could forgive people for considering it a risk they'd rather avoid.
It's trading a well-understood risk for a one that is not well-understood, hence my comment about it being irrational.