- consistent way of writing "patch" + Foam::name(id) etc.

- remove (unused) Istream constructors, prune some unused methods,
  rationalize write() vs writeDict().
  Deprecate inconsistent construction order.

- handle empty names for ".ftr" surface patches (for plain triSurface
  format) with double-quoted strings for more reliable streaming.
  Written on a single line.

  This is _backward_ compatible, but if users have been parsing these
  files manually, they will need to adjust their code.

Previously:
```
  (
  frt-fairing:001%1
  empty

  windshield:002%2
  empty
  ...
  )
```

Updated (with example handling of empty name):
```
  (
  frt-fairing:001%1 empty

  windshield:002%2 ""
  ...
  )
```

- was in surfMesh/triSurface/patches/
  now in OpenFOAM/meshes/Identifiers/surface/

This places the code closer to other identifier classes,
which aids in maintaining consistency

- make IOerror constructors explicit

ENH: shm: support for automatic faceZones

See merge request Development/openfoam!327

ENH: regIOobject store() now also registers the object

See merge request !326

Mark OLESEN authored 5 years ago and Mattijs Janssens committed 5 years ago

- previously the store() method just set the ownedByRegistry flag.
  Now ensure that it is indeed registered first.

- support register/store of tmp<> items.
  The tmp parameter is not cleared, but changed from PTR to CREF
  to allow further use.

  The implicit registration allows code simplification using the
  GeometricField::New factory method, for example.

  Old Code
  ========
      volScalarField* ptr = new volScalarField
      (
          IOobject
          (
              fieldName,
              mesh.time().timeName(),
              mesh,
              IOobject::NO_READ,
              IOobject::NO_WRITE,
              true  // Register
          ),
          mesh,
          dimless,
          zeroGradientFvPatchField<scalar>::typeName
      );
      ptr->store();

  New Code
  ========
      auto tptr = volScalarField::New
      (
          fieldName,
          mesh,
          dimless,
          zeroGradientFvPatchField<scalar>::typeName
      );
      regIOobject::store(tptr);

  or even
      regIOobject::store
      (
          volScalarField::New
          (
              fieldName,
              mesh,
              dimless,
              zeroGradientFvPatchField<scalar>::typeName
          )
      );

The collated container ('decomposedBlockData') is always binary
but the 'payload' might be ascii so use that header information
instead of the decomposeBlockData header.

TUT: misc cleanup in various tutorials

See merge request Development/openfoam!324

  BAKW: ensures bitwise backward compatibility

  BAKW: ensure bitwise backward compatibility

BUG: wrong bounding of sensitivity contituents in case of many control boxes (Fixes #1549)

Closes #1549

See merge request Development/openfoam!325

When more than one volumetric B-Splines control boxes are present, the
sensitivity constituents corresponding to the non-active design
variables were not bounded(zeroed) correctly. The resultant
sensitivities, used in the optimization, were bounded correctly, so this
was more a bug pertaining to the output file of the sensitivities rather
than a functional one.

  BAKW: ensure bitwise backward compatibility

DEFEATURE: deprecate v2f model in favour of kEpsilonPhitF

See merge request Development/openfoam!319

Kutalmış Berçin authored 5 years ago and Andrew Heather committed 5 years ago

  - kEpsilonPhitF is a kEpsilon-based model which originated
    from (Durbin, 1995)’s v2-f methodology. However, the majority of
    v2-f model variants proved to be numerically stiff for segregated
    solution algorithms due to the coupled formulations of v2 and f fields,
    particularly on wall boundaries.

    The v2-f variant (i.e. OpenFOAM’s v2f model) due to
    (Lien and Kalitzin, 2001) reformulated the original v2-f model to enable
    segregated computations; however, a number of shortcomings regarding
    the model fidelity were reported in the literature.

    To overcome the shortcomings of the v2-f methodology, the v2-f approach
    was re-evaluated by (Laurence et al., 2005) by transforming v2 scale into
    its equivalent non-dimensional form, i.e. phit, to reduce the numerical
    stiffness.
    This variant, i.e. kEpsilonPhitF, is believed to provide numerical
    robustness, and insensitivity to grid anomalies while retaining the
    theoretical model fidelity of the original v2-f model.

    Accordingly the v2f RANS model is deprecated in favour of the variant
    kEpsilonPhitF model.

BUG: continuation of updateMethods with empty activeDesignVariables (#1540)

See merge request Development/openfoam!322

Vaggelis Papoutsis authored 5 years ago and Andrew Heather committed 5 years ago

When activeDesignVariables are not set explicitly, all design variables
are treated as active. These were allocated properly when starting from
0 but not when starting from an intermediate optimisation cycle
(e.g. running 5 optimisation cycles, stopping and restarting).

TUT: added a new tutorial including the restart of an optimisation run
to help identify future regression