Handle segment writer crash

[kjnilsson]

Which could occur when a ra server deletes its own directory whilst the segment writer is flushing a mem table to disk.

Nearly impossible to create a test for this particular case but here is a stack trace from a real test deployment:

pid: <0.242.0>
registered_name: ra_log_segment_writer
exception error: no match of right hand side value {error,enoent}
  in function  ra_log_segment_writer:handle_cast/2 (src/ra_log_segment_writer.erl, line 201)
  in call from gen_server:try_handle_cast/3 (gen_server.erl, line 2371)
  in call from gen_server:handle_msg/6 (gen_server.erl, line 2433)
ancestors: [ra_log_sup,<0.231.0>,ra_systems_sup,ra_sup,<0.197.0>]
message_queue_len: 3```