[WIP] Remove Observable Helper

[PDoakORNL]

Proposed changes

Make OperatorEstBase::write virtual and override for estimators that do not simply write their data block to hdf5 actually implemented. Integration test added.

My only explanation here is that we had inadequate testing, and there is plenty more "corner" cases like this totally uncovered to be found.

This is a draft because for whatever reason I have not put together code to successfully use hdfarchive to read a previous written dataset. Anyone who can point out my issue please do so.

What type(s) of changes does this code introduce?

Delete the items that do not apply

• Bugfix
• New feature
• Testing changes (e.g. new unit/integration/performance tests)
• Documentation changes

Does this introduce a breaking change?

• No

What systems has this change been tested on?

Mac laptop

Checklist

Update the following with a yes where the items apply. If you're unsure about any of them, don't hesitate to ask. This is
simply a reminder of what we are going to look for before merging your code.

• Yes/No. This PR is up to date with current the current state of 'develop'
• Yes. Code added or changed in the PR has been clang-formatted
• Yes This PR adds tests to cover any new code, or to catch a bug that is being fixed
• Yes Documentation has been added (if appropriate)

[camelto2]

In the ObservableHelper::set_dimensions call already has an offset into the data as a function argument (the lower_bound variable in ObservableHelper). The problem with SpinDensityNew seems to be in SpinDensityNew::registerObservableEstimator. In that we register both the SpinDensity/u and SpinDensity/d to the hdf5 and set the dimensions. However, the offset is passed in as 0 in both cases, even though the data_ is twice the size to account for up and down. Thats why it writes the up twice...it started from index 0 in the underlying data for both u and d.

It seems that all of this is fixed by using oh.set_dimensions(ng, s * derived_parameters_.npoints) in that API.

Maybe I'm wrong, but if you set the correct starting index using this set_dimensions API, you don't need to make OperatorEstBase::write virtual since it is already taken care of.

[PDoakORNL]

You're right about that. But I think now a better solution would be just to drop ObservableHelper altogether its adding 0 value here.

[quantumsteve]

Can we do the cast during construction of expanded_data?

Suggested change

	std::vector<QMCT::FullPrecRealType> expanded_data(data_.size(), 0.0);
	std::copy_n(data_.begin(), data_.size(), expanded_data.begin());
	std::vector<QMCT::FullPrecRealType> expanded_data(data_.cbegin(), data_.cend());

[quantumsteve]

Suggested change

	copy(dims.begin(), dims.end(), my_dims.begin() + 1);
	std::copy(dims.begin(), dims.end(), my_dims.begin() + 1);