Update Oceananigans to 0.98 (on main, not the registered version)

[tomchor]

This version has some differences for the interface in CliMA/Oceananigans.jl#4687, an important bugfix in CliMA/Oceananigans.jl#4753, soon it'll have significant performance improvements in CliMA/Oceananigans.jl#4706

Closes #53

[tomchor]

I'm getting this error using a V100 with this new version:

ERROR: LoadError: Failed to compile PTX code (ptxas exited with code 255)
Invocation arguments: --generate-line-info --verbose --gpu-name sm_70 --output-file /glade/derecho/scratch/tomasc/tmp/jl_1ANeFK7oJe.cubin /glade/derecho/scratch/tomasc/tmp/jl_V0TgVSf6JK.ptx
ptxas fatal   : Value 'sm_70' is not defined for option 'gpu-name'
If you think this is a bug, please file an issue and attach /glade/derecho/scratch/tomasc/tmp/jl_V0TgVSf6JK.ptx
Stacktrace:
  [1] error(s::String)
    @ Base ./error.jl:35
  [2] compile(job::GPUCompiler.CompilerJob)
    @ CUDA /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/compiler/compilation.jl:356
  [3] actual_compilation(cache::Dict{Any, CUDA.CuFunction}, src::Core.MethodInstance, world::UInt64, cfg::GPUCompiler.CompilerConfig{GPUCompiler.PTXCompilerTarget, CUDA.CUDACompilerParams}, compiler::typeof(CUDA.compile), linker::typeof(CUDA.link))
    @ GPUCompiler /glade/work/tomasc/.julia/packages/GPUCompiler/Ecaql/src/execution.jl:245
  [4] cached_compilation(cache::Dict{Any, CUDA.CuFunction}, src::Core.MethodInstance, cfg::GPUCompiler.CompilerConfig{GPUCompiler.PTXCompilerTarget, CUDA.CUDACompilerParams}, compiler::Function, linker::Function)
    @ GPUCompiler /glade/work/tomasc/.julia/packages/GPUCompiler/Ecaql/src/execution.jl:159
  [5] macro expansion
    @ /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/compiler/execution.jl:373 [inlined]
  [6] macro expansion
    @ ./lock.jl:267 [inlined]
  [7] cufunction(f::typeof(CUDA.partial_mapreduce_grid), tt::Type{Tuple{typeof(identity), typeof(min), Float64, CartesianIndices{1, Tuple{Base.OneTo{Int64}}}, CartesianIndices{1, Tuple{Base.OneTo{Int64}}}, Val{true}, CUDA.CuDeviceVector{Float64, 1}, CUDA.CuDeviceVector{Float64, 1}}}; kwargs::@Kwargs{})
    @ CUDA /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/compiler/execution.jl:368
  [8] cufunction
    @ /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/compiler/execution.jl:365 [inlined]
  [9] macro expansion
    @ /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/compiler/execution.jl:112 [inlined]
 [10] mapreducedim!(f::typeof(identity), op::typeof(min), R::CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}, A::CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}; init::Float64)
    @ CUDA /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/mapreduce.jl:229
 [11] mapreducedim!
    @ /glade/work/tomasc/.julia/packages/CUDA/Wfi8S/src/mapreduce.jl:169 [inlined]
 [12] _mapreduce(f::typeof(identity), op::typeof(min), As::CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}; dims::Colon, init::Nothing)
    @ GPUArrays /glade/work/tomasc/.julia/packages/GPUArrays/u6tui/src/host/mapreduce.jl:76
 [13] _mapreduce
    @ /glade/work/tomasc/.julia/packages/GPUArrays/u6tui/src/host/mapreduce.jl:33 [inlined]
 [14] mapreduce
    @ /glade/work/tomasc/.julia/packages/GPUArrays/u6tui/src/host/mapreduce.jl:28 [inlined]
 [15] _minimum
    @ ./reducedim.jl:1017 [inlined]
 [16] _minimum
    @ ./reducedim.jl:1016 [inlined]
 [17] minimum
    @ ./reducedim.jl:1012 [inlined]
 [18] coordinate_summary
    @ /glade/work/tomasc/.julia/packages/Oceananigans/6tVZ7/src/Grids/grid_utils.jl:312 [inlined]
 [19] coordinate_summary
    @ /glade/work/tomasc/.julia/packages/Oceananigans/6tVZ7/src/Grids/vertical_discretization.jl:36 [inlined]
 [20] dimension_summary(topo::Bounded, name::String, dom::Tuple{Float64, Float64}, spacing::Oceananigans.Grids.StaticVerticalDiscretization{OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}}, pad_domain::Int64)
    @ Oceananigans.Grids /glade/work/tomasc/.julia/packages/Oceananigans/6tVZ7/src/Grids/grid_utils.jl:306
 [21] show(io::IOBuffer, grid::RectilinearGrid{Float64, Bounded, Periodic, Bounded, Oceananigans.Grids.StaticVerticalDiscretization{OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}}, Float64, Float64, OffsetArrays.OffsetVector{Float64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, OffsetArrays.OffsetVector{Float64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, GPU{CUDA.CUDAKernels.CUDABackend}}, withsummary::Bool)
    @ Oceananigans.Grids /glade/work/tomasc/.julia/packages/Oceananigans/6tVZ7/src/Grids/rectilinear_grid.jl:340
 [22] show
    @ /glade/work/tomasc/.julia/packages/Oceananigans/6tVZ7/src/Grids/rectilinear_grid.jl:326 [inlined]
 [23] print(io::IOBuffer, x::RectilinearGrid{Float64, Bounded, Periodic, Bounded, Oceananigans.Grids.StaticVerticalDiscretization{OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}, OffsetArrays.OffsetVector{Float64, CUDA.CuArray{Float64, 1, CUDA.DeviceMemory}}}, Float64, Float64, OffsetArrays.OffsetVector{Float64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, OffsetArrays.OffsetVector{Float64, StepRangeLen{Float64, Base.TwicePrecision{Float64}, Base.TwicePrecision{Float64}, Int64}}, GPU{CUDA.CUDAKernels.CUDABackend}})

But apparently no error using an A100...