release: 0.2.8

Ships the self-guard expansion, and corrects what the benchmark tables claim.
Measured with wafer, gforth and SwiftForth all native on x86-64 -- the macOS
sf64 build runs under Rosetta 2 and flatters us -- Fibonacci is 1.16x rather
than 0.83x, so sf64 still wins it and wafer takes the other four. README and
OPTIMIZATIONS now carry both tables.

crates/cli/Cargo.toml

@@ -9,7 +9,7 @@ license.workspace = true
 workspace = true
 
 [dependencies]
-wafer-core = { path = "../core", version = "0.2.7" }
+wafer-core = { path = "../core", version = "0.2.8" }
 wasmtime = { workspace = true }
 anyhow = { workspace = true }
 clap = { version = "4", features = ["derive"] }