what to do when your solar PV installer goes out of business?

You didnt mention Rapid Shutdown compliance which will become an issue as jurisdictions implement that part of the code. So far only microinverters and SolarEdge have that capability integrated. SMA has Tigo which can be installed with most string inverters.

Design and build it fit for purpose with as few components that have sufficient quality for the duty will likely produce a system with the GREATEST MTBF - that is, lower failure rate/downtime.

ore stuff, more goes wrong - but only if well designed and built in the first place.

Aside from the idea that for most users who probably look at their output about as often as they look at their electric meter - that is - not much if at all, I'd wager that a zero output system problem will probably get noticed sooner than 1 or 2 panels under producing and gesulting in a few % penalty in output, particularly after day/day out put variation due to weather.

Its' about probability. More stuff - higher probability of something going wrong.

Not everyone needs be concerned with wire length. But if my farthest panel was connected to a microinverter,
the wire loop to the meter would be 1500 feet long. Running that distance at 240VAC would require twice as
much copper as running panel 360VDC for the same losses, I would NOT call that, no difference. But that is
NOT the only issue. The PoCo loves to run the line at or above the legal maximum voltage. Any drop in the
line fed by inverters adds to that, often enough causing inverter monitors to cause overvoltage shutdown. The
DC line losses do not contribute to the effect, so it is minimized by running as much of the length as possible
with panel DC voltage.

Are string systems more fire prone, where is that reported? What about all those micros each a potential fire,
especially if not individually fused?

MPPT may be a point at any given instant, but the curve running very close to that same efficiency is
somewhat broad. I have measured enough panel operating voltages to see that. The only way to approach
20% loss in a string is with extremely mismatched panels or a fault. I doubt I exceed 2% anywhere.

If shade is a serious problem, go with micros. Here I sometimes lose a string at a time sun is fading, the
overall loss is only a few percent, and micros could only recover a fraction of that. Gaining a tiny bit more
is pointless if more is burned up in the power transmission.

In my case expansion was actually easier with a string system, because no inverters or
consolidated power wire was added. Added strings aligned to peak at different times
of day may share the same inverter(s), which also happens under cloudy skies. Micros
are essentially unable to do this, resulting in much more idle equipment. Bruce Roe