Incorrect ground truths in `test-tune.spca()`

[Max-Bladen]

🐞 Describe the bug: Errors raised in R-CMD-check. Very annoying as the values produced when these tests are run locally differ from those run via GitHub actions

---

🔍 reprex results from reproducible example including sessioninfo():

══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-tune.spca.R:15'): tune.spca works ────────────────────────────
object$choice.keepX[[1]] not equal to 25.
1/1 mismatches
[1] 35 - 25 == 10
── Failure ('test-tune.spca.R:16'): tune.spca works ────────────────────────────
object$choice.keepX[[2]] not equal to 35.
1/1 mismatches
[1] 15 - 35 == -20
── Failure ('test-tune.spca.R:41'): tune.spca works with NA input ──────────────
object$choice.keepX[[2]] not equal to 25.
1/1 mismatches
[1] 15 - 25 == -10

---

🤔 Expected behavior: No errors should be raised

---

💡 Possible solution: Either adjust and control RNG or reduce values checked against (latter is temporary fix)