ALE mass matrix update

[Chunheng Zhao]

Dear Neks,

I have some questions related to the mass matrix update in ALE, and char_conv.

1. We update bm1 and store bmlag by gengeom and geneig when igeom=2, so for igeom=1, bm1 and bmlag are just for the current mesh.
2. We evaluate the mass and other matrixes by using int_vel, which is used for RK substeps. 
3. For nbd=2, we have to evaluate p0 use u(t-2dt) and u(t-dt). While their relative mass matrix are bmlag1 and bm1. Which means, u(t-dt) is actually the current mesh solution, and u(t-2dt) is the previous mesh solution.
4. But we evaluate bmass for u(t-2dt) and use bmass(t-2dt) computed from int_vel to the RK substep. I am confused if bmass(t-2dt)==bmlag1? Or bmass(t-dt)==bmlag1? The really first procedure when ilag=2, u1(first RK)=u(t-2t)*bmlag/bmass(t-2dt). This is quite confusing, shall we just use the u(t-2t) as u1?

Thanks,

Chunheng.