There's a lot of deafness in the Solar Energy jungle, a very fertile ground for misinformation and hazardous Cheap Charlie installation.  We are simply not up to speed compared to our Cousins across the pond and down under.

Adrian as for your concern, people mostly forget that solar energy is dc, so any volts applied (even at very low level) will promote galvanic corrosion immediately  - especially worse in marine or coastal locations.

Personally I hate the MC4 system. No doubt it was sound enough when launched by the originator German Staubli gmbh.  However there are many different makers on the market and Installers can source products as little as 1/5th of the cost of the genuine article (ca £5 ea).  Apart from manufacturing quality, these look-alike connectors are not necessarily interchangeable (like our ubiquitous 3 pin uk plug).  Current an be as much as 50 Adc so when they dont make proper contact, incipient arcing can occur and over time develop into a full arc flash breakdown.  Numerous Youtube videos out there to demonstrate the consequences.

Ask for what to do - IMHO I would insist on long tails fitted to panels so there are no MC4s on the Roof (if its not there it can't go wrong).  Said tails can be routed up to the ridge and brought down into an attic void and a convenient large Marshalling cabinet.  BTW a cat walk grid along the ridge is very convenient and barely visible from ground.  I would strongly warn against any inclination to use Cheap Charlie panels as YouTube will show the effect of fires from internal short circuits.

As I have spent a year researching the topic I could ramble on for pages, so here I simple opine your question

While I don't disagree with the sentiment I would note that unless you are ordering in enough bulk to make a panel manufacturer take notice (that's MW) you aren't going to get tails long enough to run all the way to a marshalling cabinet. Therefore joints of some kind are unavoidable.

Given that the reason for taking action is sufficently high risk of faulty connectors, there's then a tradeoff between reducing the risk of water ingress into a badly made connection on top of the (comparatively fire-resistant) tiles and an electrical arcing fault in the loft space which can't be easily isolated. And, depending on the site, there may or may not be an increased risk of cable damage in the loft space.

I would also suggest that terminating onto the DC marshalling box correctly would require if anything more skill and care, and careful product selection, than the MC4s.

And with that number of roof penetrations there's also a greatly increased risk of a faulty install leading to water ingress.

I would suggest that a better approach would be to drive down the risk of faulty connectors by ensuring that they are made only by competent personnel with the correct tools under the right conditions, to heed the guidance from the IET CoP regarding connectors from different manufacturers, and to mount the connectors suitably e.g. so they're supported and not left in a pool of water. And as always avoid low / suspect quality goods.

That is not to say that I don't have dreams of doing similar with my own personal installation when I finally get to it, but that would be a bespoke arrangement for convenience for measuring and tinkering with stringing, since I have a professional stake in the sphere, and not something I would recommend to others.

While I don't disagree with the sentiment I would note that unless you are ordering in enough bulk to make a panel manufacturer take notice (that's MW) you aren't going to get tails long enough to run all the way to a marshalling cabinet. Therefore joints of some kind are unavoidable.

Given that the reason for taking action is sufficently high risk of faulty connectors, there's then a tradeoff between reducing the risk of water ingress into a badly made connection on top of the (comparatively fire-resistant) tiles and an electrical arcing fault in the loft space which can't be easily isolated. And, depending on the site, there may or may not be an increased risk of cable damage in the loft space.

I would also suggest that terminating onto the DC marshalling box correctly would require if anything more skill and care, and careful product selection, than the MC4s.

And with that number of roof penetrations there's also a greatly increased risk of a faulty install leading to water ingress.

I would suggest that a better approach would be to drive down the risk of faulty connectors by ensuring that they are made only by competent personnel with the correct tools under the right conditions, to heed the guidance from the IET CoP regarding connectors from different manufacturers, and to mount the connectors suitably e.g. so they're supported and not left in a pool of water. And as always avoid low / suspect quality goods.

That is not to say that I don't have dreams of doing similar with my own personal installation when I finally get to it, but that would be a bespoke arrangement for convenience for measuring and tinkering with stringing, since I have a professional stake in the sphere, and not something I would recommend to others.