Define symbol early enough for use in math

[sogaiu]

There was a CI error related to #203.

My current guess is that due to recent changes in hashing in janet, the order in which the key-value pairs are evaluated has changed.

So before may be:

spork/spork/math.janet

Lines 360 to 363 in 44e2601

	{:m (def m
	(/ (- (* cl sxy) (* sx sy))
	(- (* cl sxx) (* sx sx))))
	:b (- (/ sy cl) (/ (* m sx) cl))})))

worked fine because luckily the (def m ...) happened before the use of m for the value associated with :b.

This PR just moves the def form out of the struct.

[sogaiu]

I'm not entirely clear on why CI is failing.

FWIW, locally (also a Linux environment), the PR content appears to work fine, i.e. jpm test and janet bundle/test.janet lead to no errors.

As an additional data point, I get what looks similar to the CI error output if I do janet bundle/test.janet for 1bed0c9.

[bakpakin]

The CI error almost looks as though it testing the wrong version of spork. Looking into this.