internal/postgres: ensure doc HTML is valid UTF-8

We've seen documentation that is not valid UTF-8
(github.com/xuender/oil/math). That results in a DB error when written
to a TEXT column.  Make the documentation valid before writing to the
DB.

Change-Id: Ib2ccdd1dfb3b66af9d444b81216f561a8d906d8a
Reviewed-on: https://team-review.git.corp.google.com/c/golang/discovery/+/757995
Reviewed-by: Julie Qiu <julieqiu@google.com>
CI-Result: Cloud Build <devtools-proctor-result-processor@system.gserviceaccount.com>

internal/postgres/insert_module.go

diff --git a/internal/postgres/insert_module.go b/internal/postgres/insert_module.go
index 2fe2c5b..4c3e33f 100644
--- a/internal/postgres/insert_module.go
+++ b/internal/postgres/insert_module.go
@@ -274,7 +274,7 @@
 			m.ModulePath,
 			p.V1Path,
 			p.IsRedistributable,
-			p.DocumentationHTML,
+			makeValidUnicode(p.DocumentationHTML),
 			pq.Array(licenseTypes),
 			pq.Array(licensePaths),
 			p.GOOS,
@@ -465,7 +465,7 @@
 				continue
 			}
 			id := pathToID[path]
-			docValues = append(docValues, id, doc.GOOS, doc.GOARCH, doc.Synopsis, doc.HTML)
+			docValues = append(docValues, id, doc.GOOS, doc.GOARCH, doc.Synopsis, makeValidUnicode(doc.HTML))
 		}
 		docCols := []string{
 			"path_id",