New allow_error field for module system

[jgfouca]

This is necessary since implementations of environment modules are
highly variable and some systems produce stderr output even when
things are working fine.

Default this to false since we want to catch errors on standard
systems.

Test suite: code-checker, by-hand
Test baseline:
Test namelist changes:
Test status: bit for bit

Fixes #1625

User interface changes?: Y, new allow_error attribute for module systems

Update gh-pages html (Y/N)?: N

Code review: @jedwards4b

[lvankampenhout]

I guess I'm doing something wrong here. I've added allow_error to my config_machines.xml in the following way:
<module_system type="module" allow_error="true">

Setting up a case using create_test now fails with the following error message
ERROR: Command: '/usr/bin/xmllint --noout --schema /nfs/home1/kampe004/cesm/cesm_tags/cesm2_0_exp10j/cime/config/xml_schemas/config_machines.xsd /nfs/home1/kampe004/cesm/cesm_tags/cesm2_0_exp10j/cime/config/cesm/machines/config_machines.xml' failed with error '/nfs/home1/kampe004/cesm/cesm_tags/cesm2_0_exp10j/cime/config/cesm/machines/config_machines.xml:2329: element module_system: Schemas validity error : Element 'module_system', attribute 'allow_error': The attribute 'allow_error' is not allowed.

I noted that the error message references XSD file config/xml_schemas/config_machines.xsd whereas the commits in this issue carried out changes in config/xml_schemas/env_mach_specific.xsd

Any ideas?

[jedwards4b]

Yes, we need to add this field to the schema definition. The problem would have been detected earlier had full testing been done on this PR prior to merging.

[jedwards4b]

Fixed in #2654

[lvankampenhout]

Thanks Jim

[jgfouca]

@jedwards4b , does E3SM not do XSD checking on this file? We've had allow_error on for months.

[rljacob]

Its probably version 1 vs version 2 behavior.

[jedwards4b]

It seems like if you created a case on melvin, summit or summitdev it would have an error. The test is on env_mach_specific.xml rather than on the config_machines.xml file.

[jgfouca]

@jedwards4b , you'd think so, but i've created 10's of thousands of cases on melvin since this was introduced.

[jedwards4b]

I think @rljacob may be right and it's the version , but then I don't understand why @lvankampenhout ran into the problem - is he using cesm or e3sm?

[jgfouca]

@jedwards4b he may be the first CESM person to try to use it.

[jedwards4b]

ah - that makes sense then.

[lvankampenhout]

yes I'm using CESM