Splitgraph

* Fix MySQL plugin crashes on binary data types.

Full set of changes: [`v0.2.6...v0.2.7`](https://github.com/splitgraph/sgr/compare/v0.2.6...v0.2.7)

* Fix querying when there are NULLs in primary keys (https://github.com/splitgraph/sgr/pull/373)
* Data source and foreign data wrapper for querying CSV files in S3 buckets and HTTP (https://github.com/splitgraph/sgr/pull/397)
* Ctrl+C can now interrupt long-running PostgreSQL queries and stop sgr (https://github.com/splitgraph/sgr/pull/398)
* Support for updating miscellaneous repository metadata from the `sgr cloud metadata` CLI (https://github.com/splitgraph/sgr/pull/399)

Full set of changes: [`v0.2.5...v0.2.6`](https://github.com/splitgraph/sgr/compare/v0.2.5...v0.2.6)

* Fix piping CSV files from stdin (https://github.com/splitgraph/sgr/pull/350)
* Truncate commit comments if they're above the max field size (currently 4096) (https://github.com/splitgraph/sgr/pull/353)
* Add support for updating repository topics from the CLI (`sgr cloud metadata`) (https://github.com/splitgraph/sgr/pull/371)

Full set of changes: [`v0.2.4...v0.2.5`](https://github.com/splitgraph/sgr/compare/v0.2.4...v0.2.5)

* Mount handlers are now called "data sources", a generalization that will make them more pluggable
and support sources beyond FDWs. See https://github.com/splitgraph/sgr/pull/324 for more documentation and necessary steps to migrate.
* Added `sgr singer target`, a Singer-compatible target that can read [Singer tap](https://github.com/singer-io/getting-started/blob/master/docs/SPEC.md#singer-specification) output from stdin and build Splitgraph images. It's based on a fork of https://github.com/transferwise/pipelinewise-singer-python with additions that let us produce deltas and ingest them directly as Splitgraph objects.
* Support for dynamically loading plugins without specifying them in `.sgconfig`, by looking up plugins in a certain directory (see https://github.com/splitgraph/sgr/pull/329)

Full set of changes: [`v0.2.3...v0.2.4`](https://github.com/splitgraph/sgr/compare/v0.2.3...v0.2.4)

* Socrata FDW now correctly emits `IS NULL / IS NOT NULL`, same with ES (using ES query syntax).
* Fix array handling (`a IN (1,2,3)` queries get rewritten and pushed down correctly).
* Output more query information in `EXPLAIN` for Socrata/LQ.

Full set of changes: [`v0.2.2...v0.2.3`](https://github.com/splitgraph/sgr/compare/v0.2.2...v0.2.3)

* Add ability to pass extra server args to postgres_fdw (`extra_server_args`)
* Add ability to rename object files in-engine (utility function for some ingestion).
* Allow disabling `IMPORT FOREIGN SCHEMA` and passing a table schema in Postgres/MySQL FDWs.
* Add a fork (https://github.com/splitgraph/postgres-elasticsearch-fdw) of https://github.com/matthewfranglen/postgres-elasticsearch-fdw to `sgr mount`, letting others mount ES indexes. Fork changes:
* Pass qualifiers as ElasticSearch queries using the query DSL (was using the `query=...` qual as a Lucene query string, which is useless in JOINs. Now we combine both the query implied from the quals and the Lucene query string, if passed)
* Close the search context on `end_scan` (otherwise many ES queries to the FDW in a 10 minute span would cause it to error with a "too many scroll contexts" exception)
* Add EXPLAIN support (outputs the used ES query)

Full set of changes: [`v0.2.1...v0.2.2`](https://github.com/splitgraph/sgr/compare/v0.2.1...v0.2.2)