Add support for debug info.

[maleadt]

Demo with vadd from the README:

julia> ct.@device_code_tiled debuginfo=true ct.launch(vadd, grid, a, b, c, ct.Constant(tile_size))
// vadd(cuTile.TileArray{Float32, 1, cuTile.ArraySpec{1, 128, true, (0,), (16,)}()}, cuTile.TileArray{Float32, 1, cuTile.ArraySpec{1, 128, true, (0,), (16,)}()}, cuTile.TileArray{Float32, 1, cuTile.ArraySpec{1, 128, true, (0,), (16,)}()}, cuTile.Constant{Int64, 16})

cuda_tile.module @kernels {
  entry @vadd(%arg0: tile<ptr<f32>>, %arg1: tile<i32>, %arg2: tile<i32>, %arg3: tile<ptr<f32>>, %arg4: tile<i32>, %arg5: tile<i32>, %arg6: tile<ptr<f32>>, %arg7: tile<i32>, %arg8: tile<i32>) {
    %cst_16_i64 = constant <i64: 16> : tile<i64> loc(#loc)
    ...
  } loc(#loc23)
} loc(#loc)
#di_file = #cuda_tile.di_file<"REPL[3]" in "">
#di_file1 = #cuda_tile.di_file<"operations.jl" in "/home/tim/Julia/pkg/cuTile/src/language">
#di_file2 = #cuda_tile.di_file<"arithmetic.jl" in "/home/tim/Julia/pkg/cuTile/src/language">
#loc = loc(unknown)
#loc1 = loc("REPL[3]":2:0)
...
#loc22 = loc("/home/tim/Julia/pkg/cuTile/src/language/operations.jl":229:0)
#di_compile_unit = #cuda_tile.di_compile_unit<file = #di_file>
...

julia> @device_code_sass ct.launch(vadd, grid, a, b, c, ct.Constant(tile_size))
        .target sm_120a

        .elftype        @"ET_EXEC"


//--------------------- .text.vadd                --------------------------
        .section        .text.vadd,"ax",@progbits
        .align  128
        .global         vadd
        .type           vadd,@function
        .size           vadd,(.L_x_5 - vadd)
        .other          vadd,@"STO_CUDA_ENTRY STV_DEFAULT"
vadd:

.text.vadd:
; Location REPL[3]:2
        LDC R1, c[0x0][0x37c] ;
; Location /home/tim/Julia/pkg/cuTile/src/language/operations.jl:47
        S2UR UR5, SR_CTAID.X ;
; Location REPL[3]:2
        LDCU UR7, c[0x0][0x388] ;
; Location /home/tim/Julia/pkg/cuTile/src/language/operations.jl:47
        CS2R.32 R6, SR_CgaSize ;
        IMAD R6, R6, -0xa0, RZ ;
        R2UR UR6, R6 ;
        LDCU UR6, c[0x0][UR6+0x2b0] ;
; Location REPL[3]:2
        USHF.R.S32.HI UR4, URZ, 0x1f, UR7 ;
        ULEA.HI UR4, UR4, UR7, URZ, 0x4 ;
; Location /home/tim/Julia/pkg/cuTile/src/language/operations.jl:47
        UI2FP.F32.U32 UR5, UR5 ;
; Location REPL[3]:2
        ULOP3.LUT UR8, UR4, 0xfffffff0, URZ, 0xc0, !UPT ;
; Location /home/tim/Julia/pkg/cuTile/src/language/operations.jl:47
        UFMUL UR5, UR6, UR5 ;

It's ironic that cuTile.jl now has full debug info before CUDA.jl...