Replace remaining LSB protocol implementations with ordinary functions

[athos]

This is the last step of #289.

This PR adds the following implementations of LSB protocols as ordinary functions:

• reader
  • DataInput (defined in cljam.io.util.lsb.data-io)
  • InputStream (defined in cljam.io.util.lsb.io-stream)
• writer
  • OutputStream (defined in cljam.io.util.lsb.io-stream)

The cljam.io.util.lsb namespace has some other implementations as well, but they are no longer in use.

The following table shows the implementations used before and after the change:

• reader

  	before change	after change
  cljam.io.bam.core	DataInputStream (DataInput)	cljam.io.util.lsb.data-io
  cljam.io.bam.reader	DataInputStream (DataInput)	cljam.io.util.lsb.data-io
  cljam.io.bam-index.reader	FileInputStream (InputStream)	cljam.io.util.lsb.io-stream
  cljam.io.bcf.reader	BGZFInputStream (InputStream)	cljam.io.util.lsb.io-stream
  cljam.io.bigwig	RandomAccessFile (DataInput)	cljam.io.util.lsb.data-io
  cljam.io.csi	DataInputStream (DataInput)	cljam.io.util.lsb.data-io
  cljam.io.tabix	DataInputStream (DataInput)	cljam.io.util.lsb.data-io

• writer

  	before change	after change
  cljam.algo.convert	ByteBuffer	cljam.io.util.lsb.io-stream
  cljam.io.bam.encoder	DataOutputStream	cljam.io.util.lsb.io-stream
  cljam.io.bam.writer	DataOutputStream	cljam.io.util.lsb.io-stream
  cljam.io.bam-index.writer	DataOutputStream	cljam.io.util.lsb.io-stream
  cljam.io.bcf.writer	DataOutputStream	cljam.io.util.lsb.io-stream
  cljam.io.csi	DataOutputStream	cljam.io.util.lsb.io-stream

I've confirmed that this change doesn't have significant negative impact on the performance of the related readers/writers (This PR also includes a tweak of the benchmark code for the BAM reader to make it more stable and reliable by reducing extra overhead of sequence allocation in the benchmark code itself):

	before change	after change
BAM reader	5.835351 s	5.677128 s
BAM writer	12.149866 s	12.076689 s
BCF reader	6.469314 s	6.468553 s
BCF writer	1.641983 s	1.676683 s
cljam.algo.convert	182.333152 ms	180.971159 ms
BigWig reader	84.117612 μs	83.471144 μs
CSI reader	22.210736 μs	21.966407 μs
CSI writer	127.886597 μs	127.627769 μs
Tabix reader	24.434094 μs	25.902402 μs

See below for more about the benchmarking details:

benchmarking details

For those with existing benchmark code, I benchmarked them using the following lein libra commands:

target	benchmarking command
BAM reader	lein libra :only cljam.io.sam-bench/decode-bam-alignment-long-bench
BAM writer	lein libra :only cljam.io.sam-bench/encode-alignment-long-bench
BCF reader	lein libra :only cljam.io.vcf-bench/decode-variant-large-bcf-bench
BCF writer	lein libra :only cljam.io.vcf-bench/encode-variant-large-bcf-bench
cljam.algo.convert	lein libra :only cljam.algo.convert-bench

For the others, I used the following tiny benchmark code:

(require '[criterium.core :as cr]))

;; BigWig reader
(require '[cljam.io.bigwig :as bigwig])
(cr/quick-bench
 (with-open [r (bigwig/reader "test-resources/bigwig/test-bedgraph.bigWig")]
   (dorun (bigwig/read-tracks r))))

(require '[cljam.io.csi :as csi])
;; CSI reader
(cr/quick-bench (csi/read-index "test-resources/csi/test-v4_3-complex.vcf.gz.csi"))
;; CSI writer
(let [idx (csi/read-index "test-resources/csi/test-v4_3-complex.vcf.gz.csi")]
  (cr/quick-bench (csi/write-index "out.csi" idx)))

;; Tabix reader
(require '[cljam.io.tabix :as tbi])
(cr/quick-bench (tbi/read-index "test-resources/tabix/test.gtf.gz.tbi"))

[codecov]

Codecov Report

Attention: 7 lines in your changes are missing coverage. Please review.

Comparison is base (d6d1af8) 88.31% compared to head (8612caa) 87.94%.

Files	Patch %	Lines
src/cljam/io/csi.clj	73.91%	0 Missing and 6 partials ⚠️
src/cljam/io/util/lsb/io_stream.clj	98.87%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #293      +/-   ##
==========================================
- Coverage   88.31%   87.94%   -0.37%     
==========================================
  Files          81       83       +2     
  Lines        7058     7192     +134     
  Branches      488      488              
==========================================
+ Hits         6233     6325      +92     
- Misses        337      379      +42     
  Partials      488      488              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alumi]

LGTM 👍
Now that cljam.io.util.lsb is no longer used, we can either remove the ns and cut a new release 0.9.0, or deprecate functions in it and keep the version 0.8.5.
I've searched our internal repositories and found just one function depending on the ns, so I think it's almost safe to remove it. Could you tell me what you think about this?

[athos]

I’d prefer more graceful deprecation, so I've added the :deprecated attr to cljam.io.util.lsb.

I also added another commit to add notes that warn use of the new lsb namespaces to discourage undesired abuse of them from outside cljam in the future.

[niyarin]

LGTM 👍