perf: parallelize terraform module analysis

[RoseSecurity]

Why

Summary

• Parallelize Terraform module analysis across CPU cores using a bounded worker pool with goroutines, channels, and a semaphore
• Merge FindTFDirs and FindFiles into a single ScanRepository function that walks the filesystem once instead of twice
• Expand and unify the directory skip list (adding .terragrunt-cache, node_modules) so irrelevant subtrees are pruned early via filepath.SkipDir

Benchmarks

Stage	Wall time	CPU	Speedup
Before	937ms	60%	—
+ parallel workers	450ms	220%	2.1x
+ single walk & skip list	365ms	279%	2.6x

Test plan

• go build ./... passes
• go vet ./... passes
• Verified output matches pre-change results on the same repository
• Test on a large multi-module Terraform repository to validate parallel speedup scales

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}