docs: document parse behavior with non-format strings

[elliotttf]

I came across this behavior while using ptera to parse ISO strings without milliseconds. It's clear from the tests that this is the intended behavior, but I didn't notice this documented anywhere, so this PR updates the docs to cover the behavior of parse when non-format strings are in play.

Alternatively, the parse function could be updated to only handle format strings. A test to confirm the behavior can be added by applying this diff

diff --git a/parse_date_test.ts b/parse_date_test.ts
index a6729e5..0a64d2e 100644
--- a/parse_date_test.ts
+++ b/parse_date_test.ts
@@ -259,6 +259,21 @@ Deno.test("parseDateStr valid", () => {
         locale: "en",
       },
     },
+    {
+      dateStr: "2021-01-12T01:02:03Z",
+      format: "YYYY-MM-ddTHH:mm:ssZ",
+      expected: {
+        year: 2021,
+        month: 1,
+        day: 12,
+        hour: 1,
+        minute: 2,
+        second: 3,
+        millisecond: 0,
+        offsetMillisec: 0,
+        locale: "en",
+      },
+    },
   ];
   tests.forEach((t) => {
     assertEquals(parseDateStr(t.dateStr, t.format), t.expected);