update comment, explaining the reasoning behind setting a lower bound for allocated UL grants

srsenb/src/stack/mac/sched_ue_ctrl/sched_ue_cell.cc

@@ -298,7 +298,7 @@ int get_required_prb_dl(const sched_ue_cell& cell,
 
 uint32_t get_required_prb_ul(const sched_ue_cell& cell, uint32_t req_bytes)
 {
-  const static int MIN_ALLOC_BYTES = 10; /// There should be enough space for RLC header + BSR + some payload
+  const static int MIN_ALLOC_BYTES = 10;
   if (req_bytes == 0) {
     return 0;
   }
@@ -317,7 +317,9 @@ uint32_t get_required_prb_ul(const sched_ue_cell& cell, uint32_t req_bytes)
   uint32_t final_tbs = std::get<3>(ret);
   while (final_tbs < MIN_ALLOC_BYTES and req_prbs < cell.cell_cfg->nof_prb()) {
     // Note: If PHR<0 is limiting the max nof PRBs per UL grant, the UL grant may become too small to fit any
-    //       data other than headers + BSR. In this edge-case, force an increase the nof required PRBs.
+    //       data other than headers + BSR. Besides, forcing unnecessary segmentation, it may additionally
+    //       forbid the UE from fitting small RRC messages (e.g. RRCReconfComplete) in the UL grants.
+    //       To avoid TBS<10, we force an increase the nof required PRBs.
     req_prbs++;
     final_tbs = compute_tbs_approx(req_prbs);
   }