Spread is Float but spread is typically different per row of the OHLC dataframe

[FCNealVillangca]

Enhancement description

so I was looking at the Backtest class and it accepts spread: float = .0, which is as far as I know spread is different per row and it's not fixed. I'm looking forward to the spread to be either np array or a required column in data: pd.DataFrame,.

Additional info, images

this is what Mt5 return so as you notice the spread has it's own column because it's dynamically changing depending on time.

First 5 rows:
                        Open     High      Low    Close  Volume  spread  real_volume
time
2025-04-16 22:00:00  1.13864  1.14128  1.13847  1.13865    6294       8            0
2025-04-16 23:00:00  1.13866  1.13992  1.13816  1.13981    2788       8            0
2025-04-17 00:00:00  1.13980  1.13994  1.13939  1.13964     627      21            0
2025-04-17 01:00:00  1.13971  1.13997  1.13943  1.13995    1461       8            0
2025-04-17 02:00:00  1.13995  1.14093  1.13940  1.13955    2776       8            0
2025-04-17 03:00:00  1.13959  1.13969  1.13651  1.13677    6776       8            0
2025-04-17 04:00:00  1.13677  1.13692  1.13553  1.13654    6591       8            0
2025-04-17 05:00:00  1.13655  1.13819  1.13651  1.13782    3625       8            0
2025-04-17 06:00:00  1.13782  1.13808  1.13640  1.13642    3374       8            0
2025-04-17 07:00:00  1.13641  1.13699  1.13603  1.13623    2376       8            0
2025-04-17 08:00:00  1.13623  1.13712  1.13441  1.13522    3616       8            0
2025-04-17 09:00:00  1.13523  1.13711  1.13435  1.13589    6353       8            0
2025-04-17 10:00:00  1.13586  1.13789  1.13529  1.13770    8072       8            0
2025-04-17 11:00:00  1.13767  1.13885  1.13681  1.13785    7003       8            0
2025-04-17 12:00:00  1.13785  1.13808  1.13701  1.13727    5780       8            0
2025-04-17 13:00:00  1.13727  1.13766  1.13575  1.13596    7365       8            0
2025-04-17 14:00:00  1.13596  1.13701  1.13559  1.13564    5638       8            0
2025-04-17 15:00:00  1.13560  1.13680  1.13352  1.13564   12521       8            0
2025-04-17 16:00:00  1.13567  1.13776  1.13509  1.13611   12523       8            0
2025-04-17 17:00:00  1.13615  1.13857  1.13384  1.13496   10400       8            0
2025-04-17 18:00:00  1.13498  1.13766  1.13476  1.13632    7736       8            0
2025-04-17 19:00:00  1.13632  1.13712  1.13398  1.13505    8103       8            0
2025-04-17 20:00:00  1.13505  1.13538  1.13410  1.13510    4468       8            0
2025-04-17 21:00:00  1.13511  1.13752  1.13430  1.13733    5151       8            0
2025-04-17 22:00:00  1.13732  1.13760  1.13685  1.13731    3522       8            0
2025-04-17 23:00:00  1.13732  1.13766  1.13615  1.13619    1848       8            0
2025-04-18 00:00:00  1.13614  1.13688  1.13587  1.13666     334      41            0
2025-04-18 01:00:00  1.13674  1.13709  1.13656  1.13675    3927       8            0
2025-04-18 02:00:00  1.13675  1.13747  1.13657  1.13727     701       8            0
2025-04-18 03:00:00  1.13727  1.13862  1.13611  1.13740    3889       8            0

[AtomicGlance]

I opened PR #1393 with a focused implementation of per-bar spreads. It accepts scalar, 1-D array-like, and same-index Series values, validates finite non-negative rates, applies the value at the fill bar, and preserves scalar behavior. The full test module passes (83 passed, 1 skipped).