The word "concelebras" never has a long vowel in the third syllable. Its third syllable has a short vowel, but can optionally be scanned as a "heavy" syllable in poetry because the short vowel is followed by the consonant sequence "br".
This is a case of the "muta cum liquida" rule, which occurs in words where a short vowel is followed by a consonant that can be pronounced in the same syllable as a following L or R sound (you can check for this by seeing if the consonant can come before L or R at the start of a word).
Whether a syllable containing a short vowel in this kind of context scans as light or heavy is pretty freely determined by the requirements of the meter. There are some additional rules based on the structure of a word, and tendencies based on which consonants are involved, but you usually won't need them.
The word accent can never fall on the first syllable of concelebras. When the second-to-last syllable is pronounced light, which was probably the most common pronunciation, there is no question of the accent going anywhere but on the third-to-last syllable.
In a situation where the second-to-last syllable scans as heavy, it would probably be accented (but I'm not sure whether this scansion and accentuation should be counted as a poetic license). There is a relevant quote from Quintilian, Inst. Orat. I.5.28 regarding how to accent “pecudes pictaeque volucres” cited on this webpage: "Some notes on how to read Latin verse", Alatius.com, by Johan Winge. But you're unlikely to find the third syllable of concelebro scanned as heavy in a line of dactylic hexameter because there is no justification for lengthening the preceding syllable -ce-, which is only followed by a single consonant. If you search for concelebr* in the hexameter/pentameter database pedecerto, you'll see that only one result out of 33 scans with -cē- instead of -cĕ-, and it looks questionable to me (carmina epigraphica Bücheler - Lommatzsch CIL 09, 06315 = CLE 00383 = CLERegio-IV 00016 = Cholodniak 00562).