Consider falling back to non-cached behavior when redis is unavailable

[sigurdm]

Error: Bad state: StreamSink is bound to a stream

at ._StreamSinkImpl.close
at ._Socket.close
at .RespClient.close ( package:neat_cache/src/providers/resp.dart:224 )
at .RedisCacheProvider._withResp ( package:neat_cache/src/providers/redis.dart:143 )
at Future.timeout.<fn>
at ._Entry.set ( package:neat_cache/neat_cache.dart:194 )
at ._Entry.get ( package:neat_cache/neat_cache.dart:178 )
at .TaskBackend._taskResultIndex ( package:pub_dev/task/backend.dart:790 )
at .TaskBackend.gzippedTaskResult ( package:pub_dev/task/backend.dart:848 )
at .TaskBackend.dartdocFile ( package:pub_dev/task/backend.dart:903 )
at handleDartDoc.<fn> ( package:pub_dev/task/handlers.dart:142 )
at .handleDartDoc ( package:pub_dev/task/handlers.dart:141 )
at .documentationHandler ( package:pub_dev/frontend/handlers/documentation.dart:83 )

Here we could potentially go read the documentation from the bucket again, instead of returning 500.