julia 0.6 broadcast integration

[iblislin]

This work is derived from #321 (comment)

This PR make generic function do broadcast call correctly.

e.g.

julia> using MarketData

julia> f(a, b, c, d, e) = a * b + c - d + e 

julia> f.(1, 1, cl, ohlc, op[1:10])  # this was borken before this PR
10x4 TimeSeries.TimeArray{Float64,2,Date,Array{Float64,2}} 2000-01-03 to 2000-01-14 

             Close_Open_Open  Close_High_Open  Close_Low_Open  Close_Close_Open 
2000-01-03 | 112.94           105.32           116.13          105.88     
2000-01-04 | 103.5            101.13           110.56          109.25
2000-01-05 | 105.0            98.19            105.75          104.75      
2000-01-06 | 96.0             95.12            107.12          107.12
⋮
2000-01-11 | 93.75            90.31            99.19           96.94
2000-01-12 | 88.19            87.69            96.69           96.0
2000-01-13 | 97.75            93.48            99.73           95.48  
2000-01-14 | 101.44           99.19            102.06          101.0

---

Performance

The performance of current implementation is quite poor and the memory requirement is larger :(.
(I'm going to read more about performance tips. Please advise)

A test case for log.(2, cl .+ 1) which the works fine before.
Before:

julia> @benchmark log.(2, $cl .+ 1) 
BenchmarkTools.Trial:               
  memory estimate:  38.28 KiB       
  allocs estimate:  1269            
  --------------                    
  minimum time:     111.731 μs (0.00% GC)                                
  median time:      114.278 μs (0.00% GC)                                
  mean time:        118.402 μs (2.94% GC)                                
  maximum time:     2.330 ms (91.95% GC)                                 
  --------------                    
  samples:          10000           
  evals/sample:     1

After:

julia> @benchmark log.(2, $cl .+ 1) 
BenchmarkTools.Trial:               
  memory estimate:  85.19 KiB       
  allocs estimate:  2554            
  --------------                    
  minimum time:     218.412 μs (0.00% GC)                                
  median time:      224.006 μs (0.00% GC)                                
  mean time:        238.519 μs (4.23% GC)                                
  maximum time:     3.578 ms (87.72% GC)                                 
  --------------                    
  samples:          10000           
  evals/sample:     1

---

TODO

• add test cases
• improve performance

[coveralls]

Coverage decreased (-0.04%) to 85.012% when pulling 3444ef5 on iblis17:broadcast into 3de29a7 on JuliaStats:master.

[iblislin]

current patch set:

julia> @benchmark log.(2, $cl .+ 1)                                                                                                               
BenchmarkTools.Trial:               
  memory estimate:  41.47 KiB       
  allocs estimate:  1251            
  --------------                    
  minimum time:     93.868 μs (0.00% GC)                                                                                                          
  median time:      96.204 μs (0.00% GC)                                 
  mean time:        100.758 μs (3.53% GC)                                                                                                         
  maximum time:     2.244 ms (92.37% GC)                                                                                                          
  --------------                    
  samples:          10000           
  evals/sample:     1               

shell> git log | head                                                                                                                             
commit da49c48a78f633d5e56c3659d38bde1132212926                                                                                                   
Author: Iblis Lin <iblis@hs.ntnu.edu.tw>                                                                                                          
Date:   Sun Sep 3 19:19:13 2017 +0800                                                                                                             
                                                                                                                                                  
    broadcast: wip: rewrite broadcast_c in generated func                                                                                         

[coveralls]

Coverage increased (+1.1%) to 86.192% when pulling 74c66a8 on iblis17:broadcast into 3de29a7 on JuliaStats:master.

[coveralls]

Coverage increased (+0.3%) to 85.377% when pulling a1c412f on iblis17:broadcast into 3de29a7 on JuliaStats:master.

[iblislin]

I think it's ready for review

[coveralls]

Coverage increased (+0.3%) to 85.377% when pulling 31ca453 on iblis17:broadcast into 3de29a7 on JuliaStats:master.

[iblislin]

Bump?

[coveralls]

Coverage increased (+0.2%) to 86.093% when pulling 1acf2c6 on iblis17:broadcast into ac8b6cb on JuliaStats:master.

[iblislin]

I'm going to merge this PR tmr, if no objection