Speed up shell startup 9x (1.9s → 0.2s)

Lazy-load Intel oneAPI setvars.sh (~1.5s) via wrapper functions that
source the environment on first use of icc/icx/ifort/etc.

Cache all shell completion outputs to ~/.cache/bash.d/ so they are
sourced from disk instead of regenerated via subprocess on every
shell start.  Cache invalidates automatically when the tool binary
is updated.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

00-completion-cache

@@ -0,0 +1,25 @@
+# shellcheck shell=bash
+# Cache completion scripts to avoid subprocess overhead on every shell start.
+#
+# Usage:  cached_completion <tool> <generation-command...>
+# Example: cached_completion gh gh completion -s bash
+#
+# The cache file is regenerated when the tool binary is newer than the cache.
+# Cache location: ${XDG_CACHE_HOME:-$HOME/.cache}/bash.d/
+
+_COMP_CACHE="${XDG_CACHE_HOME:-$HOME/.cache}/bash.d"
+
+cached_completion() {
+    local tool="$1"; shift
+    local cache_file="${_COMP_CACHE}/${tool}.bash"
+    local tool_path
+    tool_path="$(command -v "$tool" 2>/dev/null)" || return
+
+    if [[ ! -f "$cache_file" || "$tool_path" -nt "$cache_file" ]]; then
+        mkdir -p "$_COMP_CACHE"
+        "$@" > "$cache_file" 2>/dev/null || return
+    fi
+
+    # shellcheck disable=SC1090
+    . "$cache_file"
+}