o Fix handling of the 'filter' arg (transcripts(), etc...) when the txdb has
  user seqlevels on.

o Add 'filter' arg to transcriptsBy(), exonsBy(), cdsBy(),
  fiveUTRByTranscript(), and threeUTRByTranscript(). exonsBy(), cdsBy(), and
  *UTRByTranscript() should at least support filtering by gene or transcript
  id.

o makeTxDbFromUCSC() - Address the CDS non-completeness issue.
  The status of the CDS (complete or not) can be inferred from the
  cds*Stat field (when present) of the UCSC table.

o Too many helper functions are defined and used internally to query the
  db:
  - dbEasyQuery() is defined in AnnotationDbi and GenomicFeatures with
    different definitions.
  - AnnotationDbi:::dbQuery() is the same as GenomicFeatures:::dbEasyQuery()
  - queryAnnotationDb()
  Clean this mess!

o DB schema change: Replace tx_chrom, exon_chrom, and cds_chrom columns with
  _tx_chrom_id, _exon_chrom_id, and _cds_chrom_id.

o Add the following indices to the db schema:
    CREATE INDEX splicing_tx_id ON splicing (_tx_id);
    CREATE INDEX splicing_exon_id ON splicing (_exon_id);
    CREATE INDEX splicing_cds_id ON splicing (_cds_id);
    CREATE INDEX gene_tx_id ON gene (_tx_id);

