So wot about using GPS to position for the ILS then, in the absence of radar vectors?
With a radar equipped airport, there cannot be an issue (if you told them so, etc).
With a non radar airport, the default ATC expectation will be that you will track to the beacon, perhaps go around the hold, fly outbound, and intercept the LOC. If you asked for a direct intercept, it should be OK of course, but IME they are a lot less likely to agree to it. Lydd is a classic case of this, with its silly "half-hour" DME arc. I have on occassions asked for a direct intercept when booking the ILS with a phone call but when I got there they wouldn't entertain it.
I am no expert on procedure design but there has to be a difference between a procedural NDB/VOR/etc IAP and a radar-vectorable ILS IAP. The former could potentially have its obstacle clearance designed to be assured only when you fly the full published version. The latter must be designed so that you are safe within the entire radar vectoring area, and if you do a self intercept
and obey the limits of the radar vectoring area (which is published) then you should be OK. Obviously at Lydd, Manston, Bournemouth there is nothing to hit
at the ILS platform for many miles around
but in principle this is dodgy e.g. you must not descend to the ILS platform until firstly inside the MSA (shown on the 25nm rad circle on the plate) and then secondly inside the radar vectoring area, but your GPS is unlikely to depict the boundary of the latter. That is why, in a non radar environment, it seems obvious that you should be flying the full published procedure even to a radar vectorable ILS, unless the ILS platform is above the SSA.
If it is VFR panel mount or hand held, it is a great addition to situational awareness while flying the needle as a primary indication.
A lot of people say that but no law actually says that, however.
What struck me in your comment was your point that the NDB approach can be flown within tolerance (with the error on the wrong side) and still hit terrain.
I think not. I think that the NDB procedure designers do actually know just how absolutely crappy they are (even if their paymasters never admit it) and they put in big margins. At Shoreham, provided you obey the DME stepdowns, you should be OK with a good 30 degree departure from the inbound track