Improve Kernel assemble methods in AbstractKernel_Class

[shishiousan]

When I use high-order element for space in kernel, an error message come.

code

CALL spaceFE%GetGlobalElemShapeData(elemsd=elemsd, xij=xij, geoElemSD=linElemSD)

error message

[WRONG ARGS] size(xij, 2) .NE. size(geoElemsd%N, 1)

It is because, for example in the case of Quadrangle9, size(xij, 2) has 9 but size(linElemSD%N, 1) always has 4 due to its linear interpolation setting.

My suggestion is replacing the above code with following one

      IF (SIZE(xij, 2) .EQ. SIZE(linElemSD%N, 1)) THEN
        CALL spaceFE%GetGlobalElemShapeData(elemsd=elemsd, xij=xij,  &
          & geoElemSD=linElemSD)
      ELSE
        CALL spaceFE%GetGlobalElemShapeData(elemsd=elemsd, xij=xij)
      END IF