are you using the new versions of electrum or the old ones?
if you shared the transaction (or maybe another one on testnet that looks like it if privacy is a concern) it could be analyzed better.

the only time the size is changed is when the signatures you produce end up needing an additional 00 at the beginning of their r value when they are encoded. depending on number of signatures this could increase but i don't think enough to decrease the fee by 1 sat/byte.

it does but the estimator on your online wallet should take that into consideration when showing the fee/byte values. that extra zero byte doesn't happen that often and the signers nowadays usually choose another nonce to get a lower r to avoid that extra byte.

edit: assuming this is the 1Merit address this may actually be a bug in electrum with uncompressed public keys.
if we remove the 15 extra bytes of uncompressed pubkeys (15*32=480 byte) from the size the fee turns into 7.5 sat/byte. that means the estimator messed up here not knowing the pubkey takes 65 byte not 33.

If you can live with a slightly restricted character set, you can make native segwit (bc1qxxxx) vanity addresses... even legacy compatible segwit (3xxxx) vanity addresses.

Like bc1qfeedch80zxged6apxyknugt67e47nsw00tpguu for example. Feed Ch-ildren? heh... used to be 1food.

This transaction was confirmed last night. Fees have been slowly dropping for 4.5 days now:

Don't underestimate the current mempool size though: there's 13 vMB to clear before confirming transactions under 6 sat/byte, and 61 vMB for transactions under 4 sat/byte.

Noted: boating accident. lost all bitcoins too.

Today really was a good day for those who were waiting for good opportunities to consolidate their inputs. I had a next block confirmation using a transaction fee of 6 sat/vByte. I made another consolidation at a little above 5 sats which got confirmed in an hour. I was planning to send some other less important transactions but the window has passed. Right now, you need 70-80 sats to get in the next 3-5 blocks the way it stands now.   

I got one confirmed yesterday at 10 sats per byte (very fast conf). It has 120 conformations now so only took 4 hours at most to confirm.

A couple of hours ago I got confirmed a transaction of 5 sat-vByte, the one I sent three weeks ago:

Miners are chewing their way through mempool: