I cracked, and bought one today.
Comments:-
If you turn on phantom power while recording, the whole recording digitally mutes on all channels for a few seconds - so be careful not to turn on channel 4 phantom when intending to turn on the limiter.
The limiter does appear to operate in the analogue domain, although the block diagram shows it after the A to D converter. I tested this by connecting the mic inputs to the output of a Mackie mixer, whose output was switched to mic level. Then I played some material from an iPod through the mixer into the R-44 mic inputs. As you increase the analogue gain, a little indicator to the left of the level display lights up when the preamp clips (even though the actual meter isn't going over the top, if you have the digital level fader down a bit). Turn on the limiter, and the preamp clipping light no longer comes on at the peaks, and the resulting recording when examined in Audition shows undamaged waveforms. However, the limiter is really a compressor with a threshold set at about -10dB. So, if you don't want the recording compressed significantly, you should aim to have the "forseen peaks" knocking around -12dB. Then the limiter will only cut in on the occasional "unforseen peaks", which shouldn't be too noticable. Normalise afterwards to make up the gain.
Noise levels at the mic input are good (imho). I tested this with an MS stereo pair of Sennheiser MKH series mics, first fed directly into the R-44 (with the MS "effect" on, which works very nicely), then with a Sennheiser mic preamp designed for these mics feeding the R-44 at line level. The recording was made at night in a suburban bedroom with a faintly ticking clock and distant traffic rumble, plus a bit of TV sound coming from another room. After carefully aligning the levels between the two recordings, I couldn't tell the difference nor see the difference in Audition's spectral display. As far as I am concerned, for all practical purposes, after running the preamp level at one which I know would be appropriate for recording classical chamber music, it's effectively noise free as far as I'm concerned.
I've not had a chance to check out battery life, but will do so over the next couple of nights, with a set of 2600mAh rechargable batteries, and a lithium polymer external battery pack. I did try yanking out the mains adapter plug during a recording and indeed the internal batteries take over nicely. I should be able to get some pretty prodigious running times from the external battery pack handing over to the internals, which I think should work.
Build quality is impressive - it's pretty hefty in the hand. Level and trim knobs are indeed fiddly. Markers are not written into recorded files (that's pretty lame). Display is small but very clear and easy to read from a variety of operating positions. Internal mics are quite usable if you really needed to. The scrub wheel does scrub when the transport is paused - in increments of 3/100 sec, which is where the 100th second display comes in - enabling you add a marker at exactly the right place (but more's the pity they don't show in the DAW).
Chief downside for my purposes is lack of pannable, balance controlled monitoring - and while recording you can't even control the relative channel balance as the mixer menu can't be reached. This means that once you go beyond a stereo pair, you'll need to use an external monitor mixer running off the four channel out connectors,to really judge what you're getting, or if you want to do any balanced playbacks. The monitoring provision does enable you to judge whether there's actually a problem with any of the channels, as you can go through them one by one, so if that's all you need you won't be troubled by what troubles me.
I'm hoping to try it for real at a string quartet recital on Sunday.