Turn off computed caching in GraphNode/Link to fix slow reaction. Implement isTransferred in GraphLink.
Store: load modules automatically, rename module files

Store/data: fix dataset validation #27, inform store/interaction about removed nodes, minor refactor
Store/interaction: respond to removed nodes, change unselect mutation

GraphNode: fix bundle count.
DatasetEditor: nicer message

/spend 2h

NodeDetailCard: add unbundle option, disable remove btn.

Pass graph visuals as a prop down to GraphNode/link to allow managing it from single place (if there will be better zooming someday). Fix sticky paws on group nodes. Linting

Implement aggregation behaviour for bundling. Allow bundling on arbitrary node type, it only worked on regular nodes

CI: deploy to different folders according to branch (master/deploy in analysis-vis/, others in analysis-vis/{branchName}/

GraphNode/Link: do not emit events, instead access store/interaction directly; this removes unnecessary indirection and the complexity of NetworkGraph comp.
GraphNode: open detail on doubleclick, add debouncing

Store/data: document tag data structure, some changes in aggreagation algorithms for bundling (WIP), changes in isLinkSimilar

GraphLink: compute ifTransferred
GraphNode: make label semitransparent, fix vm.$destroy watch
Harry dataset: fix some mistakes
Remove old datasets

Store/data: add actions for aliasing and bundling, rename aggr. mutations, remove deprecated code, remove isUserDefined when creating aggregations, add tests for bundling (not passing), remove isUserDefined.

NodeSelection: add test UI for aggregations.
GraphNode: add bundle visuals, get bundle size from store.
NetworkGraph: tweak settings, renamed store actions.

Remove unused test
Added new random dataset for performance testing

Grouping/ungrouping algorithm seems finished, done minor refactor. Finished test "complex 1". Fix GraphNode: transition on removed node allowed user to interact with it, triggering error when getting node from store/data. #13

Grouping would like a refactor.
Also, the randomized test over big dataset is failing so there may be a corner case yet to be discovered. Not sure if bug, it may be also triggered by a bug in dataset.