Plain text support for seqcli ingest

[nblumhardt]

• Multi-line STDIN frame reader
• seqcli ingest plain text by default (opt-in to JSON with --json)
• By default, all non-empty lines read will be timestamped and emitted as events
  • Any trailing lines indented with whitespace will be treated as continuations
• Apply patterns to input frames
• Unmatched (remainder) text is attached in a "remainder" event property
• Plain text files can be ingested using -i <filename>
• Extend --invalid-data= to include report as an option, so that "orphan"/un-parseable lines can be captured
• Pattern language for token matching, e.g. --pattern="[{@t:iso8601ts} {@l}] ({SourceContext}) {@m:*}{:n}{@x:lines}"
• --test flag that prints JSON to STDOUT instead of sending events to Seq (could perhaps be called something like --echo?)
• List the built-in matchers in the README
• Sanity check the (future) story for optional ({:?), parameterized (digit(4)) and regex matchers

[nblumhardt]

I've moved the remaining TODOs to #29 - once #28 is in I think we should merge this as a decent first cut.