Add release notes for 8.0.0

[elliottt]

Update the release notes to include changes since 7.0.0.

[jameysharp]

Wow, we have some cool features in this release!

[jameysharp]

• This is probably important for embedders with custom signal handlers: x64: Take SIGFPE signals for divide traps #6026
• @fitzgen maybe should write up the measured performance improvements from cranelift-wasm: Add a bounds-checking optimization for dynamic memories and guard pages #6031
• I don't know what this is but it sounds worth mentioning: Add a --dynamic-memory-reserved-for-growth CLI flag #5980
• New profiling "perfmap" mode and changed CLI flags for profiling: Add support for generating perf maps for simple perf profiling #6030

Many of this round of Cranelift changes should have an impact (hopefully positive!) on Wasmtime:

• We got a bunch of support for making use of more specialized instructions such as AVX on x86 where available (x64: Add a smattering of lowerings for shuffle specializations #5930, x64: Add more support for more AVX instructions #5931, x64: Optimize store-of-extract-lane-0 #5924, x64: Improve memory support in {insert,extract}lane #5982, x64: Add shuffle specialization for palignr #5999, x64: Add lea-based lowering for iadd #5986, x64: Add support for the pblendw instruction #6023, x64: Improve codegen for splats #6025, x64: Fix vbroadcastss with AVX2 and without AVX #6060, x64: emit_cmp: use x64_test for comparisons with 0 #6086, x64: Add AVX support for some more float-related instructions #6092)
• Fixed bugs and suboptimal codegen in aarch64 (aarch64: Add specialized shuffle lowerings #5977, Cranelift: aarch64: fix undefined dest reg in f32x4.splat case. #5987, aarch64: Use VCodeConstant for f64/v128 constants #5997, aarch64: Specialize constant vector shifts #5976, Cranelift: ISLE: aarch64: fix imm12_from_negated_value for i32, i16 #6078)
• Fixed bugs and suboptimal codegen in riscv64 (Codegen fix atomic_cas with samll types on riscv64 #5919, Codegen fix fcvt_from_sint.f32 with small types on riscv64. #5964, riscv64: Fix underflow in call relocation handling #5951, riscv64: Fix regalloc panic with bor+bnot on floats #5857, riscv64: Improve ctz/clz/cls codegen #5854, riscv64: Add Zba extension instructions #6087)
• We generate more cache-friendly code for traps now (Add a MachBuffer::defer_trap method #6011)
• We have a bunch of new machine-independent optimizations (Allow hoisting vconst instructions out of loops #5909, cranelift: simplify ineg(ineg(x)) to x #6033, cranelift: simplify fneg(fneg(x)) to x #6034, cranelift: simplify x-x to 0 #6032, cranelift: simplify icmp against UMAX/SMIN/SMAX #6037, craneleft: cancel ineg when args to imul #6053, cranelift: rewrite x*-1 to ineg(x) #6052, cranelift: rewrite iabs(ineg(x)) and iabs(iabs(x)) #6072, ISLE: rewrite and/or of icmp #6095, ISLE: rewrite loose inequalities to strict inequalities and strict inequalities to equalities #6130)

[saulecabrera]

The initial integration between Wasmtime and Winch was merged yesterday too. It's behind a cargo feature, so I'm not sure if this one is applicable for the release notes. Nevertheless I decided to bring it up just in case it is.

[alexcrichton]

Thanks! Mind adding a few more PRs too?

• Add Clone for component::InstancePre #5966
• Add a --dynamic-memory-reserved-for-growth CLI flag #5980
• Validate faulting addresses are valid to fault on #6028