docs/index.html: precise system bandwidth citation, honest noise-figure caveat

- System bandwidth row: cited TS 38.104 Table 5.3.2-1 specifically
  (verified: N_RB=51 at 20 MHz / 30 kHz SCS, matching
  nr_simulator/constants.py's N_RB=51 exactly).
- UE noise figure row: added an explicit caveat that the 7 dB value's
  TR 38.913 citation is a typical evaluation assumption, not verified
  against a specific table -- no primary-source access to confirm
  further; stated honestly rather than left implying false precision.

Closes out the two remaining open items from the 3GPP compliance table
review. README.md has no equivalent rows (confirmed, unchanged).

Author: karrthikmohank

docs/index.html

@@ -537,10 +537,10 @@ <h2 class="section-title">3GPP compliance</h2>
     <tr><td>HARQ RTT</td><td>4 slots (SCS 30 kHz → 2 ms) — simulator's full Tx-to-retransmit-eligible cycle; matches real k1 (feedback timing) default, not real systems' ~8-slot RTT (no separate scheduling/processing gap modeled)</td><td>TS 38.321 (k1 default); RTT itself is implementation-defined, not spec-fixed</td></tr>
     <tr><td>BLER target</td><td>10% (eMBB initial transmission)</td><td>TS 38.214 §5.1</td></tr>
     <tr><td>Carrier frequency</td><td>3.5 GHz (n78 band)</td><td>TS 38.104 Table 5.2-1</td></tr>
-    <tr><td>System bandwidth</td><td>20 MHz → 51 RBs (SCS 30 kHz)</td><td>TS 38.104</td></tr>
+    <tr><td>System bandwidth</td><td>20 MHz → 51 RBs (SCS 30 kHz)</td><td>TS 38.104 Table 5.3.2-1</td></tr>
     <tr><td>Inter-site distance</td><td>500 m (UMa macro)</td><td>TR 38.913</td></tr>
     <tr><td>gNB TX power</td><td>46 dBm</td><td>TS 38.104</td></tr>
-    <tr><td>UE noise figure</td><td>7 dB</td><td>TR 38.913</td></tr>
+    <tr><td>UE noise figure</td><td>7 dB</td><td>TR 38.913 (typical evaluation-assumption value; not verified against a specific table)</td></tr>
   </table>
 </div>
 </section>