Where is the $1m challenge test? There isn't one.
Given that the rules and FAQ and a list of people who have taken it (and failed) can be found on this very site, that's a very odd claim.
What is the threshold for proof?
Obviously that depends to some extent on the claim being tested, but it's usual to set up test conditions that ensure that the chances of guessing right are calculable, and then set the success criteria to be significantly better than that chance result. 1 in 1000 is usual for the preliminary test, i.e. the claimant must produce results that have only a one in a thousand chance or better of being obtained by luck alone.
JE is expected to work with mr Zwinge to develope it?
With JREF. Of course. How else could a test protocol that's acceptable to both him and JREF be arrived at?
For the kind of claim Edwards makes, a test protocol like the following might be suitable:
1. There are five subjects, each of whom has a recently deceased close relative. Each in turn sits close to JE but behind a screen. He is not allowed to question them directly. [Both precautions required to eliminate cold reading]
2. For each subject, JE writes down everything the spirits tell him about the subject and/or their deceased relative. Each account is identified with a number.
3. After the final reading, five copies of the five readings are made and given to the five subjects. Each picks out the reading which they think is theirs. [Note: they are
not given their reading alone and asked how accurate it is. See
the Forer Effect ].
4. The number of the reading picked by each subject is compared to the number of the reading that is actually theirs. Note that there's a 1 in 5 chance of each subject picking their own reading by chance, but a much lower chance of all five picking their own reading.
5. The test is run however many times is required for JE to beat the set success criteria.
ETA: As an example of the sort of negotiation that might go on before the test protocol was agreed: suppose Edwards objected that he doesn't always get information for a particular subject. The test protocol would be adjusted so that extra subjects are on standby, and Edwards would be allowed to pass on any particular subject for whom he was unable to produce a reading.