feature: display export limits and optimize export query

[ClemDoum]

PR description

Fixes #151
Fixes #152

Addressed both issues in the same PR. Optimizing the export query implied some API changes and hence a frontend update, I made the most of this update to also display the limit used when exporting:

trimmed.mov

Query optimization

Updated export queries from

MATCH (node)
OPTIONAL MATCH (node)-[r]-(other)
RETURN *
LIMIT 1000

to

MATCH (doc:Document)
WITH doc
LIMIT 1000
ORDER BY doc.path ASC
OPTIONAL MATCH (doc)-[rel:APPEARS_IN|SENT|RECEIVED]-(ne:NamedEntity)
RETURN apoc.coll.toSet(collect(doc) + collect(ne) + collect(rel)) AS values

Optimizations consisted in:

• specifying nodes labels and relationships types to speedup path matching
• applying on the number of match nodes (earlier in the query)
• deduplicating exports nodes and relationships

Changes

src (Java)

Changed

• refactored the DumpQuery to be less generic and reflect more closely the type of dump queries executed in practice. Previously the DumpQuery was very close the a Neo4jUtils.Query (match path + where + order by + limit), now the dump query is mean to initialized a single query restrincting the documents nodes to be dumped. The rest of the query (optional match of named entities, deduplication of the return values, applying limits) is computed in a rigid manner while calling asValidated(defaultLimit)
• exposed the export document node limit to the frontend through GET /graphs/dump/node-limit
• lazyly retrieve enterprise support as well export document node limit from the Python app

Removed

• removed unsued sorted graph dump API

plugins/neo4j-graph-widget (frontend)

Added

• added an overlay to explain graph limits and prevent user from closing download tab
• updated the extension state to store the export document node limit (retrieved from the backend)

Changed

• updated calls to the backend to the new API

neo4j-app (Python)

Added

• added AppConfig.neo4j_app_max_dumped_documents and AppConfig.neo4j_export_batch_size configuration attributes to size exports
• added a AppConfig.supports_neo4j_parallel_runtime attribute to indicate whether the neo4j instance supports parallel runtime

Changed

• optimized default export query and export configuration used in dump_graph, as well as updated the function to use a document node limit and forwarded configuration export batch size
• update the AppConfig to compute both enterprise and parallel runtime support in with_neo4j_support. Refactored the dependency codebase to compute this support when the neo4j driver is ready and updated the FastAPI app configuration with the updated configuration (see dependencies.py)

Fixed

• fixed a bug when logging progress in an empty project

neo4j script

Changed

• updated neo4j docker test image to the most recent 4.x (4.4.29)

[ClemDoum]

Log was remove on purpose here, the elapsed time here can't be correclty computed as we're streaming, this returns right away

[caro3801]

I reviewed the UI : the comments are non blocking for the next steps. Looks nice!

[caro3801]

In props, for object, a good practice is to put a default value if its not required:

 props: {
    widget: {
      type: Object,
      default:()=>{}
    }
},

if required:

 props: {
    widget: {
      type: Object,
      required:true
    }
},