Move theta input and change error check

[atb1995]

theta is changed from an optional argument in ThetaMethod to required and limited to 0 < theta < 1. ImplicitMidpoint is changed to be called TrapeziumRule. Note, in future we could add the actual Implicit Method in future.

[tommbendall]

The code changes are exactly right so as it stands this could be merged onto main. But...

1. Could you write a short description of the PR (under the "conversation" tab)
2. Do we want to think about renaming this? I've always wanted to clarify the difference between the following two schemes:

q^{n+1} = q^n + 0.5*dt*F(q^n) + 0.5*dt*F(q^{n+1})

and

q^{n+1} = q^n + dt*F(0.5*[q^n+q^{n+1}])

Since this one is the first, I had wondered if this should be called something like the TrapezoidalThetaMethod? What do you think?

[atb1995]

I've changed the name of ImplicitMidpoint

[tommbendall]

I'm afraid the branch is still failing! I also wondered if we should open an issue to implement GeneralisedMidpoint and ImplicitMidpoint discretisations?

[atb1995]

Yep I missed an ImplicitMidpoint and had some trailing whitespace! I think that would be good, although I think it will be tied in to having implicit multistage schemes with a butcher tableau (after explicit is done). I could still create an issue?

[tommbendall]

Looks good now, thanks!