Move away from Akka HTTP in favor of some more lightweight library

[bhermann]

Is your feature request related to a problem? Please describe.
The use of Akka HTTP currently prevents us of compiling the CLI down to native code using GraalVM. Despite the fact that the CLI currently works fine, it could be much faster which helps in automating query tasks.

Describe the solution you'd like
Use a light weight http client that can be used in a GraalVM build.

Describe alternatives you've considered
none

Additional context
none

[hariharansrc]

Change DelphiCli.scala based on suggestion oracle/graal#370

[hariharansrc]

Using GraalVM-1.0.0-rc10 for --allow-incomplete-classpath option

[hariharansrc]

Override default java options in sbt using set command.

set javaOptions := Seq("-Djava.lib.path=/usr/lib/jvm/default-java/lib/","-Djavax.net.ssl.trustStore=/usr/lib/jvm/default-java/lib/security/cacerts")

[hariharansrc]

Using environmental variable

• JAVA_LIB_PATH
• JAVA_TRUSTSTORE

[hariharansrc]

• Run graalvm-native-image:packageBin to package
• Execute commands using ./delphi-cli search "[dstore_1 (opcode:72)]<1"

[hariharansrc]

Default delphi client results

 Performance counter stats for './delphi-cli search [dstore_1 (opcode:72)]<1' (10 runs):

       6023.415951      task-clock (msec)         #    2.029 CPUs utilized            ( +-  0.51% )
             5,450      context-switches          #    0.905 K/sec                    ( +-  1.16% )
               488      cpu-migrations            #    0.081 K/sec                    ( +-  2.80% )
            48,341      page-faults               #    0.008 M/sec                    ( +-  0.21% )
   15,39,63,38,853      cycles                    #    2.556 GHz                      ( +-  0.50% )  (49.83%)
   14,25,93,68,246      instructions              #    0.93  insn per cycle           ( +-  0.29% )  (62.45%)
    2,74,66,57,930      branches                  #  455.997 M/sec                    ( +-  0.32% )  (62.33%)
      10,80,12,738      branch-misses             #    3.93% of all branches          ( +-  0.27% )  (62.56%)
    4,28,59,17,118      L1-dcache-loads           #  711.543 M/sec                    ( +-  0.59% )  (62.60%)
      28,20,08,742      L1-dcache-load-misses     #    6.58% of all L1-dcache hits    ( +-  0.43% )  (62.69%)
       6,17,23,160      LLC-loads                 #   10.247 M/sec                    ( +-  0.67% )  (50.09%)
       1,44,40,681      LLC-load-misses           #   23.40% of all LL-cache hits     ( +-  1.01% )  (49.90%)

            2.9691 +- 0.0206 seconds time elapsed  ( +-  0.69% )

[hariharansrc]

Native delphi cli results

Performance counter stats for './delphi-cli search [dstore_1 (opcode:72)]<1' (10 runs):

         47.173724      task-clock (msec)         #    0.127 CPUs utilized            ( +-  6.67% )
                18      context-switches          #    0.371 K/sec                    ( +-  7.44% )
                 2      cpu-migrations            #    0.032 K/sec                    ( +- 31.82% )
             4,170      page-faults               #    0.088 M/sec                    ( +-  0.09% )
       6,75,64,042      cycles                    #    1.432 GHz                      ( +-  7.88% )  (45.05%)
      12,07,66,606      instructions              #    1.79  insn per cycle           ( +-  5.98% )  (61.50%)
       2,18,85,776      branches                  #  463.940 M/sec                    ( +-  3.97% )  (67.20%)
          1,73,236      branch-misses             #    0.79% of all branches          ( +- 10.09% )  (68.53%)
       1,98,86,279      L1-dcache-loads           #  421.554 M/sec                    ( +-  4.30% )  (69.32%)
          9,45,189      L1-dcache-load-misses     #    4.75% of all L1-dcache hits    ( +-  2.51% )  (65.94%)
          1,27,342      LLC-loads                 #    2.699 M/sec                    ( +- 10.39% )  (43.12%)
            56,160      LLC-load-misses           #   44.10% of all LL-cache hits     ( +- 14.75% )  (40.84%)

            0.3726 +- 0.0318 seconds time elapsed  ( +-  8.54% )