Hiya Bob,
Sorry for not replying to your quesions on my blog - I've been away at a conference for the past few days, sorry!
If you haven't done so already, I'd definitely recommend posting to the JeeLabs forum, e.g. this thread:
http://forum.jeelabs.net/comment/9316Well done for taking the brave step to trying to use a software-configurable radio for this stuff: sounds like fun! Unfortunately I have zero experience with software radio so I probably wont be much help...
I'm afraid I'd have to agree that your data doesn't look like Current Cost data. As far as I know, all Current Cost data starts with 55 2D D4 (the 55 is the preamble, the 2DD4 is the sync word). And yes, you're right, the only valid bit pairs in a Manchester-encoded signal should be 01 or 10 (so any pairs of 11 or 00 are illegal and suggest something is broken).
I've had a quick tinker with your waveform and can't make heads nor tails of it, sorry. My only suggestion would be to find out precisely what your Current Cost TX is sending over the air by sniffing FSK data directly from the FSK input pin on the Current Cost transmitter's RFM02 module (but be very careful to isolate your computer from the transmitter; and I'd recommend taking apart a battery powered TX rather than an IAM). You might be able to use the audio-input on a sound card to sample from the transmitter's FSK pin (after isolating appropriately of course... maybe using an audio transformer?) I think the Current Cost transmitter sends a sequence of logical zeros and ones to the transmitter over the FSK input pin; ready for the transmitter to create an FSK signal.
Good luck - let us know how you get on!