Allocations when multi-threading

When activate multithreading by `FFTW.set_num_threads()`, it become slower and allocate more.
The provider here is the default "fftw".
But when using "mkl", it just use multithreading automatically depending on the problem size and allocation is 0.
Here are some benchmarks
```
data1=rand(ComplexF64,2^12)
p1=plan_fft!(data1)
```
```
julia> p1=plan_fft!(data1)
FFTW in-place forward plan for 4096-element array of ComplexF64
(dft-ct-dit/32
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dft-directbuf/130-128-x32 "n1fv_128_avx2"))

julia> @benchmark $p1*$data1
BenchmarkTools.Trial: 10000 samples with 1 evaluation.
Range (min … max): 21.800 μs … 68.700 μs ┊ GC (min … max): 0.00% … 0.00%
Time (median): 23.100 μs ┊ GC (median): 0.00%
Time (mean ± σ): 23.189 μs ± 1.335 μs ┊ GC (mean ± σ): 0.00% ± 0.00%

▁ ▁█
▂▂▂▄▄▅▅█▆███▆▃▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▁▂▂▂▂▂▂ ▃
21.8 μs Histogram: frequency by time 29.3 μs <

Memory estimate: 0 bytes, allocs estimate: 0.

julia> FFTW.set_num_threads(6)

julia> p1=plan_fft!(data1)
FFTW in-place forward plan for 4096-element array of ComplexF64
(dft-thr-ct-dit-x6/32
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dftw-direct-32/8 "t3fv_32_avx2_128")
(dft-directbuf/130-128-x32 "n1fv_128_avx2"))

julia> @benchmark $p1*$data1
BenchmarkTools.Trial: 10000 samples with 1 evaluation.
Range (min … max): 23.900 μs … 300.000 μs ┊ GC (min … max): 0.00% … 0.00%
Time (median): 48.000 μs ┊ GC (median): 0.00%
Time (mean ± σ): 50.904 μs ± 12.291 μs ┊ GC (mean ± σ): 0.00% ± 0.00%

▁▁▂▄▆█▆▄▂ ▁ ▂▁▁
▂▁▂▂▂▂▂▂▂▃▃▃▃▅▇▇███████████████▇▆▅▅▅▅▅▅▅▄▄▃▃▃▃▃▃▃▃▃▃▃▃▃▃▃▂▂▂ ▄
23.9 μs Histogram: frequency by time 83.5 μs <

Memory estimate: 4.41 KiB, allocs estimate: 53.
```
And
```
julia> versioninfo()
Julia Version 1.7.2
Commit bf53498635 (2022-02-06 15:21 UTC)
Platform Info:
OS: Windows (x86_64-w64-mingw32)
CPU: AMD Ryzen 5 2600X Six-Core Processor
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-12.0.1 (ORCJIT, znver1)
Environment:
JULIA_NUM_THREADS = 6
JULIA_PKG_SERVER = https://mirrors.tuna.tsinghua.edu.cn/julia
JULIA_EDITOR = code
```