* CompositionOperator and some refactoring

* Added SumOperator and CompositionOperator

added domain_geometry and optional range_geometry as parameter of Operator. These are saved in _domain_geometry and _range_geometry.
Updated all operator and tests to take notice of this change.

* fighting with Composition

* fix direct and adjoint for CompositeOperator

* fix composition Operator

* remove target OUTPUT to trigger build

* added unit tests

* add test for ZeroOperator

* fixes

* add numba

* removed hard coded path

* removed comments

* fix direct/adjoint with out parameter

* removed __rmul__

* use calculate_norm inherited from LinearOperator

* removed hard coded version tag

* removed cached string version

* use add_custom_target instead add_custom_command

* fix limits of docstring

* add test for IndicatorBox

* axpby as concrete method in DataContainer and BlockDataContainer

* fixed axpby and added unittest

* PDHG to use axpby

* pass num_threads to axpby

* void commit

* add seed to random in test

* NUM_THREADS can be imported from ccpi.utilities

* added test to axpby with num_threads

* Allows user to set number of threads used by openMP in C library gradient operator. Changed to release build flags

* closes #477

* added test function for c lib thread deployment

* improved thread scaling for neumann algoritims

* removed unnecessary thread sync

* reverts omp number of threads at the end of the c function call

Co-authored-by: Edoardo Pasca <edo.paskino@gmail.com>

* add force recalculation of norm

closes #394
better handling and documentation of Operator norm method.

* added docstring and test for dot_test

* add reduction min and max

closes #441

* add spacing

* fix func style

* update pnorm not quite right

* add test for SumFunction

* passed test for SumFunction

* passed test for SumFunction

* passed test for SumFunction

* rename Norm2Sq to LeastSquares

* rename Norm2Sq to LeastSquares

* delete files and move to Function.py

* fix accuracy convex conjugate

* comment proximal conjugate, have generic expression

* remove some imports

* remove imports and comment gradient, proximal conjugate , rmul

* remove imports and comment prox conj

* fix init

* fix out gradient

* fix init

* SumFunction full test

* rename Norm2Sq

* rename Norm2Sq

* change kwargs in KL, not ready yet

* change def of KL cases

* change kwargs in KL, not ready yet

* LinearOperatorMatrix FISTA cvx is not working

* change name in KL

* minor fix

* BlockDataContainer should give a geometry when shape is the same

* change convex conj when we have BDC

* remove space

* fix out for prox

* rename x to b

* add SumFunctionScalar test

* fix convex conj of constant

* fix convex conj of constant

* minor fix

* fix con conj

* add new tests

* KL not ready yet

* add docs, not ready yet

* fix documentation

* fix docs

* fix doc scaled function

* fix doc translate function

* fix modules path for documentation

* minor doc fix

* fix doc prox conj for KL

* fix vector geometry in vectordata

* fix direct/adjpint LinearOperator

* add test LinearOperatorMatrix

* fix doc and check methods for functionOperaCom

* add new test for FunctionOperCom with LinearOperMatr

* get vector geometry in vector data

* minor changes and docs

* fix docs

* check Lips cons

* rename Norm2Sq to LeastSquares for docs

* fix KL, docs, background term, conve conj

* add KL test

* fix KL

* minor fix doc

* add back prox conj work with smoothTV

* smooth TV not ready yet

* add prox conj back

* minor change

* no change

* fixing docs, not ready yet

* uncomment prox_conj for ScaledFunction

* fix doc for BlockFunction

* add smoothTV not ready yet

* no change

* fix doc and uncomment prox conj

* remove unicode literals

* sphnix is a nightmare

* fixing ADMM

* remove conv conj, prox, prox conj from smooth TV

* merge edo's docs

* fix docs for L2Norm after edo's PR

* fix docs

* remove ADMM

* add docs for Function module

* remove unicode_literals from future imports

https://stackoverflow.com/questions/7602171/unicode-error-unicodeescape-codec-cant-decode-bytes-string-with-u

* remove ADMM

* removed unicode literals from future imports

* fix init for functions

* remove main

* add explcit KL prox conj, does not work with generic prox conj of function

* add elif pass for vector geometry

* fixes from comments on the PR

* minor changes

* fix mul in Function and Identity in TranslateFunction

* remove from call

* remove BlockGeometr from BDC

* remove min, max

* cleaned Function

* add test for constant function

* remove smooth mixed L21 Norm

* fixed fill method and test of SumFunction

* Functions dev (#479)

* v19.10 docs (#467)

updated docstrings and documentation

* add Gemma in the authors

* update docs

* Installs with make install (#478)

* closes #477

* install from build dir

Co-authored-by: Vaggelis Papoutsellis <22398586+epapoutsellis@users.noreply.github.com>

updated docstrings and documentation

C library implemented with optimised axpy fucntions and gradient operator in c