> Basically would I stand a better chance of finding a way-marked Stile a km away with the 64 compared to the other two units?
Realistically, the precision with which rights of way and tracks are marked on a map, even a 1:25k map, is far poorer than the accuracy you can usually expect with any GPS receiver using the Coarse Acquisition (C/A or civilian) service, barring receiver perturbations. So finding that stile (not usually marked on maps...) will still need you to find the physical stile. Which can sometimes be a challenge in overgrown hedgerows...
A track depicted as 1mm wide on a map would be 25m wide on the ground using a 1:25k map, and 50m wide on a 1:50k map.
Receiver sensitivity is a double-edged sword. On the one hand, it allows you to pick up weak signals. On the other hand, it allows you to pick up weak signals. Weak signals can be caused by attenuation of the line-of-sight signal (e.g by tree cover), or by reflection of the line-of-sight signal from surrounding objects (e.g. tree trunks and branches). So, whilst a higher sensitivity receiver might continue to track under thick tree canopy, you should expect the fix to be degraded due to multipath and reflections.
The more satellites you can see, the better your solution ought to be, with a law of diminishing returns. So, being able to see GLONASS should give your a marginally improved fix, but it might only give you 3m average error as opposed to 5m (for example). It is unlikely to improve your Time To First Fix (from a cold start), since the two systems are separate, and you will still need to acquire the ephemerides for both constellations before the receiver can start calculating fixes. A receiver with Predictive Ephemeris capability should allow you to increase the lifetime of the stored ephemerides, and so you should usually only need a warm start, admittedly, with some degradation in position until the true ephemerides can be received.
The antenna type is often determined by the physical design of the receiver box, rather than performance. For a dedicated GPS receiver (as opposed to a phone), the chances are that the antenna performance will be good enough, and not really worth you bothering about. It's useful to know where the antenna this 'up' is, though, so you can position the receiver so that it gets the best view of the sky (rather than having half of its beam profile pointing at the ground).
For mapping; if you want to find stiles, you'll need a map depicting field boundaries. So, for the UK, that means 1:25k OS maps, as 1:50k maps don't depict field boundaries.
I know your stile comment was just an example of the accuracy you were hoping for, but it serves as a useful way of thinking around what your requirements are ;-)
For the mapping side, I'm afraid I can't help, as I don't have experience of commercial mapping receivers; I use a GPS PDA with full UK 1:50k and 1:25k mapping that fits on a 16GB card, and a Hudl running OruxMaps that will cache or download OS tiles from Bing (no longer available due to legal pressure from the OS). I also have a full set of UK mapping for Locus on the Hudl, but that takes just under 30GB. I prefer OruxMaps for user experience.
What might also be interesting to you is the navigation guidance the receiver gives you; again, I can't help here, as my GPS PDA merely plots my position on the map, and route following is up to me.
Can routes be prepared on the device, and, if so, how easy is it to do that?
How easy is it to transfer routes & logs to and from a computer?
How easy is it to manage routes on the device?
Does it offer folders to allow routes to be organised?
Does it allow enough characters in waypoint names for them to be usable?
Does it re-name your carefully named waypoints in routes?
How well does it cope with routes that double-back on themselves (for instance, when automatically advancing to the next waypoint)?
Can it distinguish between waypoints and route points (this can be a solution to the above; a genuine waypoint at the far point of a double-back route, with routepoints following the meander to and from that waypoint). I'm not even sure if GPX files understand this distinction, although one might choose to write a GPX file as a set of waypoints, between which are route or track segments.
