Much longer precompilation on Julia 0.7

[YingboMa]

julia> @time using OrdinaryDiffEq
[ Info: Recompiling stale cache file /home/scheme/.julia/compiled/v0.7/OrdinaryDiffEq/DlSv.ji for module OrdinaryDiffEq
184.484367 seconds (7.75 M allocations: 493.164 MiB, 0.16% gc time)

julia> versioninfo()
Julia Version 0.7.0-beta2.61
Commit 1d3c9c8 (2018-07-21 02:46 UTC)
Platform Info:
  OS: Linux (x86_64-pc-linux-gnu)
  CPU: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.0 (ORCJIT, skylake)
Environment:
  JULIA_PKG3_PRECOMPILE = 1

(v0.7) pkg> status
    Status `~/.julia/environments/v0.7/Project.toml`
  [c52e3926] Atom v0.7.0
  [6e4b80f9] BenchmarkTools v0.3.2
  [34da2185] Compat v1.0.0
  [3a865a2d] CuArrays v0.6.2
  [a0b5b9ef] Cxx v0.2.0+ #master (https://github.com/Keno/Cxx.jl.git)
  [bcd4f6db] DelayDiffEq v4.2.1+ [`~/.julia/dev/DelayDiffEq`]
  [2b5f629d] DiffEqBase v4.13.0+ [`~/.julia/dev/DiffEqBase`]
  [eb300fae] DiffEqBiological v3.1.0+ [`~/.julia/dev/DiffEqBiological`]
  [459566f4] DiffEqCallbacks v2.0.0
  [f3b72e0c] DiffEqDevTools v2.2.1+ [`~/.julia/dev/DiffEqDevTools`]
  [c894b116] DiffEqJump v5.0.0+ [`~/.julia/dev/DiffEqJump`]
  [78ddff82] DiffEqMonteCarlo v0.13.0+ #32b39be (https://github.com/JuliaDiffEq/DiffEqMonteCarlo.jl.git)
  [77a26b50] DiffEqNoiseProcess v2.3.0+ [`~/.julia/dev/DiffEqNoiseProcess`]
  [9fdde737] DiffEqOperators v3.2.0
  [a077e3f3] DiffEqProblemLibrary v3.6.0+ [`~/.julia/dev/DiffEqProblemLibrary`]
  [aaf54ef3] DistributedArrays v0.4.0+ [`~/.julia/dev/DistributedArrays`]
  [f6369f11] ForwardDiff v0.7.5
  [28b8d3ca] GR v0.31.0
  [e5e0dc1b] Juno v0.5.0
  [46d2c3a1] MuladdMacro v0.2.0
  [1dea7af3] OrdinaryDiffEq v4.2.0+ [`~/.julia/dev/OrdinaryDiffEq`]
  [65888b18] ParameterizedFunctions v3.7.1
  [149e707d] PkgDev v0.2.1
  [91a5bcdd] Plots v0.17.3
  [e6cf234a] RandomNumbers v0.2.0 #master (https://github.com/sunoru/RandomNumbers.jl.git)
  [731186ca] RecursiveArrayTools v0.16.3+ [`~/.julia/dev/RecursiveArrayTools`]
  [295af30f] Revise v0.5.4
  [90137ffa] StaticArrays v0.8.2
  [789caeaf] StochasticDiffEq v5.2.3+ [`~/.julia/dev/StochasticDiffEq`]
  [c3572dad] Sundials v2.1.0

julia> @time using OrdinaryDiffEq
INFO: Recompiling stale cache file /home/scheme/.julia/lib/v0.6/OrdinaryDiffEq.ji for module OrdinaryDiffEq.
 29.084996 seconds (5.12 M allocations: 280.087 MiB, 0.73% gc time)

julia> versioninfo()
Julia Version 0.6.5-pre.0
Commit 7711781* (2018-07-10 00:08 UTC)
Platform Info:
  OS: Linux (x86_64-pc-linux-gnu)
  CPU: Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT NO_AFFINITY HASWELL MAX_THREADS=16)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.9.1 (ORCJIT, skylake)

julia> Pkg.status("OrdinaryDiffEq")
 - OrdinaryDiffEq                3.20.0             c41ad05e (dirty)

I obtianed this timing by

1. typing using OrdinaryDiffEq in the REPL
2. making a minor change in OrdinaryDiffEq
3. @time using OrdinaryDiffEq.

[JeffBezanson]

This is spending most of its time collecting stacktraces while throwing MatchErrors. The @muladd macro seems to be incredibly inefficient.