Merge 1.0.0 beta.10.3 proposed update into 1.0.0

[lifflander]

Fixes #1186, #1194

[cz4rs]

ToDo:

• Build on develop or 1.0.0 fails with Intel 19 compilers #728 - in progress, contains some bits from Add general scoping for reductions #796 (like detail::ReduceStamp)
• Active messenger fails when sending over 2 GiB of data due to MPI limitations #1111
• Calling vt::initialize with no arguments silently makes VT fail #1163
• Phase assertion fails during restart #1183

Skipped in this release (to avoid further delay): #1173

[cz4rs]

• builds pass cleanly at 7290bec (skipping quality review):
  

• merge commits are missing, but the history is cleaned-up and linear

• I am currently working on backporting Build on develop or 1.0.0 fails with Intel 19 compilers #728 - main problem is that there are some bits from Add general scoping for reductions #796 involved (overall, I am trying to backport Phil's solution while maintaning the reduce interface from 1.0.0)

[lifflander]

• builds pass cleanly at 7290bec](7290bec) (skipping quality review):
• merge commits are missing, but the history is cleaned-up and linear
• I am currently working on backporting Build on develop or 1.0.0 fails with Intel 19 compilers #728 - main problem is that there are some bits from Add general scoping for reductions #796 involved (overall, I am trying to backport Phil's solution while maintaning the reduce interface from 1.0.0)

Looks good so far. Let me know if you need any help from me. The reduce interface did change a lot on develop, but I think the changes Phil made should generally be applicable to 1.0.0.

[cz4rs]

pushed single commit from #908 to fix another Intel 19 issue (95ff3ea), it might be necessary to include complete solution

ToDo:

• add nvcc builds to CI and fix any remaining issues
• add Failures when compiling with nvcc on develop #908 to the release looks like 95ff3ea is enough

[cz4rs]

@lifflander

• I have added up-to-date build definitions for nvcc
• all the essential builds are passing (including intel 19, nvidia 10.1, nvidia 11.0)

[PhilMiller]

The code changes look as expected, and EMPIRE tests pass.

[PhilMiller]

The code changes look as expected, and EMPIRE tests pass.

Those tests were with GCC 7.2.0. I'm building and testing with Cuda 9.2 now.

[PhilMiller]

The code changes look as expected, and EMPIRE tests pass.

Those tests were with GCC 7.2.0. I'm building and testing with Cuda 9.2 now.

And, they passed.

I can merge this, unless there's something else that needs to happen with it first