Fix indent in Analysis Engine docs

docs/Analysis_Engine.md

@@ -103,11 +103,11 @@ Explanation of fields (including some optional fields not present in the above q
    * `allowMoves (list of dicts)`: Optional. Same as `avoidMoves` except prohibits all moves EXCEPT the moves specified. Currently, the list of dicts must also be length 1.
    * `overrideSettings (object)`: Optional. Specify any number of `"paramName":value` entries in this object to override those params from command line `CONFIG_FILE` for this query. Most search parameters can be overriden: `cpuctExploration`, `winLossUtilityFactor`, etc. Some notable parameters include:
       * `playoutDoublingAdvantage (float)`. A value of PDA from -3 to 3 will adjust KataGo's evaluation to assume that the opponent is NOT of equal strength/compte, but rather that the current player has 2^(PDA) times as many playouts as the opponent. Dynamic versions of this are used to significant effect in handicap games in GTP mode, see [GTP example config](../cpp/configs/gtp_example.cfg).
-        * `wideRootNoise (float)`. See documentation for this parameter in [the example config](../cpp/configs/analysis_example.cfg)
-        * `ignorePreRootHistory (boolean)`. Whether to ignore pre-root history during analysis.
-        * `antiMirror (boolean)`. Whether to enable anti-mirror play during analysis. Off by default. Will probably result in biased and nonsensical winrates and other analysis values, but moves may detect and crudely respond to mirror play.
-        * `rootNumSymmetriesToSample (int from 1 to 8)`. How many of the 8 possible random symmetries to evaluate the neural net with and average. Defaults to 1, but if you set this to 2, or 8 you might get a slightly higher-quality policy at the root due to noise reduction.
-        * `humanSLProfile (string)`. Set the human-like play that KataGo should imitate. Requires that a human SL model like `b18c384nbt-humanv0.bin.gz` is being used, typically via the command line parameter `-human-model`. Available profiles include:
+      * `wideRootNoise (float)`. See documentation for this parameter in [the example config](../cpp/configs/analysis_example.cfg)
+      * `ignorePreRootHistory (boolean)`. Whether to ignore pre-root history during analysis.
+      * `antiMirror (boolean)`. Whether to enable anti-mirror play during analysis. Off by default. Will probably result in biased and nonsensical winrates and other analysis values, but moves may detect and crudely respond to mirror play.
+      * `rootNumSymmetriesToSample (int from 1 to 8)`. How many of the 8 possible random symmetries to evaluate the neural net with and average. Defaults to 1, but if you set this to 2, or 8 you might get a slightly higher-quality policy at the root due to noise reduction.
+      * `humanSLProfile (string)`. Set the human-like play that KataGo should imitate. Requires that a human SL model like `b18c384nbt-humanv0.bin.gz` is being used, typically via the command line parameter `-human-model`. Available profiles include:
            * `preaz_20k` through `preaz_9d`: Imitate human players of the given rank. (based on 2016 pre-AlphaZero opening style).
            * `rank_20k` through `rank_9d`: Imitate human players of the given rank (modern opening style).
            * `preaz_{BR}_{WR}` or `rank_{BR}_{WR}`: Same, but predict how black with the rank BR and white with the rank WR would play against each other, *knowing* that the other player is stronger/weaker than them. Warning: for rank differences > 9 ranks, or drastically mis-matched to the handicap used in the game, this may be out of distribution due to lack of training data and the model might not behave well! Experiment with care.