Fix for upstream projects using GNU make

Fixed

• Fixes for GNU Make builds

Workaround for NAG 7.1 (7010) and CI updates

Changed

• Minimum version of yaFyaml is now v1.0.1; contains workarounds for NAG 7.1 (7110)
• Updated GitHub Actions
  • OSs
    • Remove macos-10.15
    • Add ubuntu-22.04 and macos-12
  • Compilers
  • Removed gfortran-8
  • Added gfortran-11
  • Added gfortran-12 (for ubuntu-22.04)

Fixed

• Workarounds in tests for NAG 7.1 (7110)

Improved workaround for GFortran string handling

Fixed

• Introduced workaround for gFortran when combining strings with unlimited polymorphic entities. Users may be able to avoid using a String() wrapper for args to logging commands now.
• Cleaned up some stray debugging prints
• Fixed misspelling of SUCCESS

Updating to new yaFyaml interfaces requires 1.0-beta7

Changed

• Updated to new version (1.0-beta7) of yaFyaml which involves new interfaces.

Added support for NVHPC

[1.7.0] - 2022-03-24

Added

• Add NVHPC.cmake and PGI.cmake files for NVHPC support

Fixed

• Implemented workaround for gfortran 11.2 that was breaking on
  a surprisingly simple call.

Changed

• upgraded to use new gFTL2 interfaces for containers.

Apache 2.0 License

Merge pull request #62 from Goddard-Fortran-Ecosystem/hotfix/license

Introducing Apache 2.0 license.

Updated to use yaFyaml 1.0

No change to pFlogger interfaces.

CMake Namespace support and return codes

• Now provides CMake namespace. Upstream projects should link with PFLOGGER::pflogger
• All procedures with potentially failing branches are now covered with optional return code dummy arguments.

Bugfix for unit tests

Fixed one more test that was missed in previous update. (Was using -f filter to focus on some failing tests. Blinded to thore tests.)

Changes to memory management of handlers

Changed

• Recent addition of basic_config() (see previous release)
  introduced an ambiguity for memory management. The underlying
  design from Python is based upon garbage collection and reference
  counting. To solve this problem there the user can now either
  specify a HandlerVector or a HandlerPtrVector (but not both) If the
  former, pFlogger will copy the handlers into its internal storage.
  If the latter, pFlogger will only retain a pointer reference to the
  user actual argument. In that case, the user is repsonsible for
  maintaining the handlers until pFlogger is finalized and ensuring
  that any locks are freed.

• The argument to Logger::add_handle() must now have the TARGET
  attribute. pFlogger will only maintain a pointer to that target.
  This is to be consistent with the above strategy.

Fixed

• Implemented workaround for ifort 18.0.5 that became necessary after
  the changes above. Only impacted unit tests.