Time interval Dirichlete BC

[wenqing]

Added time interval Dirichlete BC with a new keyword of $TIME_INTERVAL where one can specify a Dirichlete BC existing in a time interval

Example

 $TIME_INTERVAL
  0.0 100.0
 $TIME_INTERVAL
  100.0 4.320000e+003

Also see a modified benchmark: https://github.com/ufz/ogs5-benchmarks/blob/master/H/h_frac.bc

[norihiro-w]

There is already $TIME_CONTROLLED_ACTIVE option in BC file. Have you tried it?

[wenqing]

@norihiro-w I see that there exists the similar option. I never tried that one. The advantage of that one by use a curve is that for the same geometry, only one set of BC nodal variables is allocated. With the one in the current PR, BCs with the same geometry but with different time intervals need more memory. If you prefer the existing one, I will close this PR.

[norihiro-w]

@wenqing It's not clear to me what is advantage of this PR over the existing one. I think we shouldn't have duplicated implementations if purposes of the both are the same. It would be nicer if you can improve the existing code or completely replace it with the new one.

[wenqing]

@norihiro-w Thanks for the suggestion. I will think about it.

[wenqing]

@norihiro-w Removed $TIME_CONTROLLED_ACTIVE. I want to merge it soon.

[norihiro-w]

@wenqing still it's not clear to me what is advantage of this PR over the existing one. For instance, with the existing one, you can set multiple intervals (e.g. active if 0<t<10, 20<t<30) to the same boundary condition.

[wenqing]

@norihiro-w Added a new commit. With the change, the code can handle multi-intervals in a BC. Also see a modified benchmark: https://github.com/ufz/ogs5-benchmarks/blob/master/H/h_frac.bc

[norihiro-w]

@wenqing i'm okay with this PR after you add an error message for $TIME_CONTROLLED_ACTIVE. It would be nice if you fix the syntax later.

[wenqing]

162 - WX-M/3D_oedometer_mohr_coulomb (Failed). However this benchmark ran well by manually running ctest on the server.