I thought that the original reason for moving from springs to foam seating was the fire risk on seats with springs was a lot higher than the fire-retarded foam. These appear to be made with small pocket springs in these - what changed?

Did you really think that? Or (as I once read somewhere) is it April 1st already ??

That Railway Gazette explains it:

So no need for spring steel to be more flammable, or have a greater combustion energy, than foam.

By having an RSSB^▸ (Rail Safety and Standards Board) account.

A comfort score is Dimensional features + Pad & Back Rest thickness + Compression.  Maximum score 100

Dimensional features includes:
a) seat height (target 440 mm ±10 mm above the floor)
b) Seat depth (good practice value to be 435 mm ±10 mm)
c) Seat width and gap between armrests (highest attainable score  for 460mm - 503mm with or without armrests, but a without score is handicapped)
d) Backrest width (highest attainable score for >549mm)
e) Armrest height (highest attainable score for fully adjustable, second highest 230mm - 250mm if non adjustable)
f) Underside of headrest to seat [distance from the seat's sitting surface to the passenger's acromion] (highest attainable score for 660mm - 680mm)
g) Angle of seat (highest attainable score for -6° to -9°)
h) Angle between seat and back (highest attainable score for 100° - 105°)
i) Legroom (highest attainable score for >766mm) [legroom is measured from the back of the seat to the back of the seat in front]
j) Bay seating arrangement (higest attainable score for > 1531mm) [measured from back of seat to the back of the seat opposite]
k) Clearance under tablet (650 - 690mm)
l) Tablet depth (151 - 200mm)

With regards to the seat pad, some choice figures are:
"It is good practice for the minimum seat pad thickness to be 50 mm."
"The seat pad's target minimum compression is 40% of the overall seat pad thickness." at 500N
"The maximum allowable compression is 70% of the overall seat pad thickness." at 1000N
"The minimum target is 5% or less of deformation after 50,000 cycles."

Metal springs obviously don't burn, from what I have read it seems to have been to do with funnel effect, basically a fire would grow in a draft along the void.  Filling with solid foam meant no void so was assumed safer.

Although what I looked at related to aircraft seats, what was subsequently realised was the foam has gaps, it doesn't totally fill the space (slots are left for usb and entertainment cables etc).  And of some trains there are power sockets.   So presumably whilst old risk of dropped cigarette has gone, there is a new risk of electrical fire.

Historically the idea of foam seem to be a bit of a disjointed policy that dates back to cigarettes being smoked, as the foam was often installed in vehicles fitted with cables that gave off toxic fumes from their insulation if there was a fire.  

It seems there is a realisation that fire rarely kills from burning, but toxic fumes incapacitate people and bodies get discovered unburnt, but dead from suffocation whilst unconscious.  There was a reference to a minor train fire in 1950s which although fire was put out some ladies had died in a compartment apparently smiling.  It was eventually discovered smouldering varnished panels had given off toxic cellulose fumes that had killed them.