Standardize Variable*

[Affie]

WIP

	VariableDFG	VariableCompute
✓	id::Union{UUID, Nothing}	id::Union{UUID, Nothing}	REMOVE
✓	label::Symbol	label::Symbol	label::Symbol
✓	tags::Vector{Symbol}	tags::Set{Symbol}	tags::Set{Symbol}
☐	timestamp::ZonedDateTime	timestamp::ZonedDateTime	timestamp::TimeDateZone
☐	nstime::String	nstime::Nanosecond	REMOVE
✓	ppes::Vector{MeanMaxPPE}	ppeDict::Dict{Symbol, AbstractPointParametricEst}	REMOVE
✓	blobEntries::Vector{Blobentry}	dataDict::Dict{Symbol, Blobentry}	blobentries::Blobentries
✓	metadata::String	smallData::Dict{Symbol, MetadataTypes}	bloblets::Bloblets
✓	solvable::Int	solvable::Base.RefValue{Int}	solvable::Base.RefValue{Int}
✓	solverData::Vector{PackedState}	solverDataDict::Dict{Symbol, VariableState{T, P, N}}	states::States
✓	variableType::String	-	type::TypeMetadata
✓	_version::VersionNumber	-

Notes:

• Consolidate VariableDFG and VariableCompute
• timestamp only, see Change to VariableDFG.missionnanosec alongside .timestamp #1087
• smallData -> bloblets
• solverDataDict-> states
• dataDict -> blobentries
• nstime different uses:
  • as 0-999_999 nanosecond part of time stamp now in timestamp,
  • as steady timestamp now in steadytime
• see pr for some progress

[Affie]

I'm now considering dropping the type parameters completely from VariableDFG and perhaps State as well. That will depend if the state containers remains a concrete type. If not, there is no use in having dynamic dispatch twice and State.belief is the only field that supports different types and dynamic dispatch can be isolated to this one field. (similar can be done for factors: {T <: AbstractObservation, N})