LILACSMOKE: Call case.load_env() before building the atm driver

[billsacks]

Description of changes

In LILACSMOKE test: Call case.load_env() before building the atm driver. This is needed for the atm driver build to use the correct module environment.

Specific notes

Things were working for me, and apparently for @ekluzek and @negin513 , without this change. But this is probably because our default module environment was similar to cime's cheyenne_intel environment. @glemieux was running into trouble because he has intel17 loaded in his default environment (whereas the environment loaded by cime uses intel19). This should fix that issue.

Contributors other than yourself, if any: none

CTSM Issues Fixed (include github issue #): none

Are answers expected to change (and if so in what way)? Answers may change for the LILACSMOKE test, if baselines were generated with a slightly different compiler version than the one loaded by cime. However, I do not expect answer changes.

Any User Interface Changes (namelist or namelist defaults changes)? No

Testing performed, if any:

• I first reproduced the error reported in Merge ctsm1.0.dev105 tag into fates_main_api  #1137 (comment) by changing my environment to match @glemieux 's. Specifically, I replaced my .profile with his, and replaced the contents of my .lmod.d directory with his. So this used intel17. This led to the same issue with the LILACSMOKE test that he reported.
• I then reran the LILACSMOKE test using @glemieux 's environment but with the fix in this PR; now it passed.
• I also ran the LILACSMOKE test using my environment (which uses intel 19.0.5) with this fix; it also passed.

[billsacks]

@ekluzek although this isn't critical like the last LILAC-related PR (it only impacts testing, not what a typical user will experience), the impact is so limited that I feel it's safe to bring it to master without any additional testing. (I have run the one test that it impacts.) Are you okay with this? If so, I'll bring it to master.